Can multi-stakeholder forums influence good governance in communal forest management? Lessons from two case studies in Ethiopia

• MSFs may improve participation, facilitate multi-sectoral approaches, and create opportunities to address power imbalances among stakeholders to communal forests. • Factors including MSF’s inclusiveness, the power relations between stakeholders, enabling joint planning and implementation, determine the effectiveness of MSFs and the credibility of its outcomes among stakeholders. • Gender inequity has a negative impact on the achievement of good governance in communal forests. • MSFs could embrace a mix of top-down and bottom-up approaches for an increased alignment and impact of efforts to address constraints to good governance in communal forests at different levels. • Government could play a central role in strengthening the capacities of MSFs towards improving the governance of communal forests.


INTRODUCTION
Good governance refers to a process in which different stakeholders engage and participate to make decisions that affect their livelihoods in an inclusive, transparent, and accountable manner (Gisselquist 2012). Good governance could be associated with whether attention is given to the diverse experiences and interactions of different categories of stakeholders in everyday life (Cornwall 2003). In this paper, we define good governance as a decision-making process in which diverse groups of stakeholders with varied interests, capacities, and power relations collaborate to achieve the sustainable management of communal forests (e.g. Yami et al. 2018). Communal forests, in relation to the Ethiopian cases studied here, refer to forest resources managed by local communities with support of local administrative bodies. Local communities manage communal forests using bylaws that define the users and their access to resources, mobilize and facilitate collective action for preventing degradation, and facilitate equal benefit sharing among users (Yami et al. 2013).
The notion of good governance has dominated the development discourse in Africa since the World Bank included it as a criterion for development assistance in the 1980s (Doornbos 2001). The concept has entertained support for its positive effect in reducing the lengthy and bureaucratic approaches of governments, thereby improving the efficiency of public services (Rogers and Hall 2003). However, some criticise the way good governance has been imposed in a dogmatic approach by international donors and the lack of clarity on the standards of its modalities (e.g. Harrison 2005, IDS 2010. Valid concerns critique that the term is overused and is little understood among governments and development partners, making its implementation problematic.
Nevertheless, the concept has drawn the attention of governments and other stakeholders in the management of protected areas and natural resources (e.g. Brown et al. 2002). In particular, governments' recognition of the demands of stakeholders such as local communities and Non-governmental Organizations (NGOs) for inclusive decision-making processes in forest management led to greater attention to participatory approaches where decisions are made with the involvement of different stakeholders . Participatory approaches may engage stakeholders at different levels including those closer to the ground, and allow them to negotiate their different interests and to collaborate during the decisions' implementation (Armitage et al. 2012, Hauck et al. 2015. Participatory approaches have contributed to increased public involvement in the design and implementation of interventions in the forestry sector, thereby stimulating good governance of forests (Bayley andFrench 2008, Mollick et al. 2018). Although the notion of good governance is not a silver bullet to solve the complex problems local communities face in rural Africa, its key principles -ensuring inclusiveness, equity, accountability, and transparency -makes its implementation important in communal forests (Yami et al. 2009(Yami et al. , 2018. In particular, communal forests in Ethiopia tend to lack inclusive decision-making processes and their sustainable management is hampered by power imbalances among stakeholders (Mekuria et al. 2019(Mekuria et al. , 2021. Multi-stakeholder forums (MSFs) are understood as participatory processes that aim to bring together all relevant stakeholders in communication and/or decision-making processes on natural resource management (Fenta andAssefa 2009, Bekele et al. 2015). MSFs are seen as promoting good governance by encouraging active stakeholder participation and by building trust and shared understanding among different stakeholder groups in forest management (Faysse 2006, Obeng et al. 2014. MSFs often involve capacity development interventions with the aim of building its stakeholders' skills in negotiation and advocacy (Obeng et al. 2014). Such interventions could empower representatives of different stakeholder groups such as women and civil society organisations (CSOs), and strengthen their negotiation skills in voicing their concerns and interests in communal forest management. This could in turn strengthen the MSFs' contributions to broader participation and improved fairness in decisionmaking processes (Fenta and Assefa 2009).
However, research reveals that participatory approaches do not always lead to positive governance outcomes in communal forests. This is partly due to corruption and the misuse of power by the authorities at local levels (Lund andTreue 2008, Dawson et al. 2018). For instance, elite capture of the benefits from Participatory Forest Management (PFM) interventions in the Eastern Arc Mountains of Tanzania constrained the effectiveness of participatory processes (Vyamana 2009). Furthermore, Ravikumar et al. (2018) noted that effective coordination among different sectors in addressing deforestation in Peru, Indonesia, and Mexico were constrained by political issues.
Similarly, in some developing countries the use of participatory processes that did not consider the socio-cultural context of the area they were set up to address led to limited consideration of the interests and concerns of indigenous peoples and other marginalized groups in Reducing Emissions from Deforestation and Forest Degradation (REDD+) and other climate-smart interventions (e.g. Schroeder 2010, Wang 2016. Accordingly, some authors argue that the realization of good governance in forest management depends on whether participatory processes pay attention to the active involvement of stakeholders in decision-making processes, power relations, land tenure systems, and the diverse interests of stakeholders, among others (e.g. Fenta andAssefa 2009, Ratner et al. 2017).
Hence, whether the participation offered by an MSF results in improved governance depends on multiple factors, such as power relations between stakeholders and whether they enable joint planning and implementation. In past decades, efforts to achieve good governance in forest management across countries in Africa and Asia were constrained by the limited understanding of MSF organizers on the power dynamics among stakeholders and how this impacted the way they work together towards a shared goal (e.g. Brouwer et al. 2013, Ratner et al. 2017, Bhattarai et al. 2018. Similarly, in the Madre de Dios region of Peru, Rodriguez-Ward et al. (2018) found that power imbalances and frictions within jurisdictions hindered the active participation of NGOs and indigenous peoples in a REDD+ MSF.
At times, an MSF's official status in the bureaucratic system could constrain its ability to operate as an independent entity, resulting in the failure to deliver both its promised outputs and inclusiveness (Faysse 2006, Pattberg andWiderberg 2016). For instance, Sahide and Giessen (2015) found that, in Indonesia, the National Forestry Council, which was established as an MSF, is a legal partner of the Ministry of Forestry in the development and implementation of policies. This MSF could be vulnerable to its control by government bodies, which would aggravate the power imbalances and interfere with the negotiation of interests among stakeholders (Hemmati 2012).
In Ethiopia, the failure of top-down and centralized interventions has drawn the attention of the government and development partners towards initiatives aimed at enhancing good governance in the management of communal forests (e.g. Lemenih et al. 2014). The government of Ethiopia has recently implemented participatory approaches that allow local authorities and communities to participate in decisionmaking over the management of communal forests, such as exclosures (Yami et al. 2013). Interest in using participatory approaches such as PFM has also increased, with the overall goal of government and development partners achieving good governance in Ethiopia's communal forests. These approaches are characterized by inclusive decision-making processes and equitable benefit-sharing among different stakeholders .
There are cases where participatory approaches have contributed to improvements in the overall condition of forests and the livelihoods of local communities in Ethiopia (e.g. Kidu et al. 2017). Yet, concerns remain on whether such efforts promote good governance in communal forests (e.g. Tadesse et al. 2017). For example, Kemerink-Seyoum et al. (2018) questioned the transparency of the criteria and basis for selecting community representatives in the Bale and Humbo communal forests MSF. They assert that socio-political relations influenced the selection of local community representatives, and also in defining the MSF members. Tekalign et al. (2015) also revealed that PFM did not result in equitable benefit sharing among local communities in Bonga, in southwestern Ethiopia. Kassa et al. (2017) emphasized that the lack of ownership of processes and outcomes by local communities, and the lack of clear benefit sharing mechanisms limit the success of forest rehabilitation interventions such as exclosures and PFM interventions.
Addressing concerns over good governance in Ethiopia is critical for three reasons. First, the high level of local community dependence on forest resources for income and livelihoods raises concerns over the distribution of benefits . Second, the land tenure system often fails to respect and enforce the rights of stakeholders over forests . And thirdly, the prevalent use of top-down approaches in decisionmaking processes with regards to the use and management of forest resources constrains the active participation of stakeholders . These three reasons indicate that putting mechanisms in place to enhance collaboration among stakeholders and improve governance of natural resources is crucial for better sustaining the processes and outcomes in the long term.
Past studies analyzed the governance challenges in the Ethiopian forestry sector, including policy development (Ayana et al. 2013, Ariti et al. 2018; equity of benefit sharing (Yami et al. 2013); stakeholder participation, and the efficiency and effectiveness of PFM cooperatives in communal forest management (e.g. Hailemariam et al. 2015). Studies noted the need for improved communication and information exchange among local stakeholders and government authorities on communal forests' technical and political issues, as well as for more effective communal forest management (Ameha et al. 2014. The evidence also demonstrates that the forestry sector in Ethiopia exhibited constraints to good governance with regards to institutional capacity and inclusiveness in decision-making processes (e.g. Lemenih et al. 2014, Kidu et al. 2017. Accordingly, there is a recent interest in implementing MSFs to fill the institutional gap by bringing together stakeholders with local and technical knowledge, and political and financial capacities (Bekele et al. 2015).
Past studies have shown that MSFs in Ethiopia have, to a limited extent, improved the interactions of NGOs with the government in policy development related to natural resource biodiversity conservation, and development of land-use plans across Ethiopia. Research sought to identify case studies that provided comparable insights on the role of MSFs in achieving sustainable land use and land-use change.
The Bale and Jamma-Urji MSFs were selected using the following criteria: they included a forum for in-person interactions; included different types of actors -at least one government and one non-government local actor; were organized at the subnational level; sought to address unsustainable land or resource challenges; and had been meeting for at least a year. The Bale MSF was perceived by participants of the scoping study as a successful MSF, whereas the Jamma-Urji MSF was perceived as one that had failed. Both MSFs were established in Oromia regional state, Ethiopia, operated at sub-national levels, and shared similar politico-administrative contexts (Table 1).

Bale
The Bale eco-region is located in the Bale zone, Oromia regional state and consists of 16 weredas 1 (Figure 1). The ecoregion covers a total land area of approximately 38,036 km 2 with altitude ranging between of 550 to 4,377 masl. The population of the eco-region is estimated at 3.3 million people (Source: Support for Horn of Africa Resilience-Bale Eco-Region, SHARE-BER, project). While Oromo is the dominant ethnic group, other ethnic groups such as Amhara and Somali also inhabit the eco-region.
The SHARE-BER project was a three-year project (2014-2017) funded by the European Union (EU). It intended to conserve biodiversity, ecosystem functions and services in the management (Ariti et al. 2018). However, little is known on the results of using MSFs to influence good governance in managing communal forests. In particular, there is insufficient comparative research that examines if and how MSFs have improved the governance of communal forests. This paper fills that gap by comparing two MSFs, Bale and Jamma-Urji MSFs, in Oromia, Ethiopia, using the Q-methodology and follow-up in-depth interviews with the MSFs' organizers and participants. The paper attempts to answer how MSFs influence good governance in communal forests by looking at their influence on inclusiveness, ownership, equity and transparency in decision-making processes.

METHODS AND STUDY AREA
The research project defined MSFs as "purposefully organized interactive processes that bring together stakeholders to participate in dialogue, decision-making and/or implementation regarding actions seeking to address a problem they hold in common or to achieve a goal for their common benefit" (Sarmiento Barletti et al. 2020:2). The study sites were selected after a scoping study was conducted on the experience of MSFs in the forestry sector in Ethiopia. The scoping study assessed eight MSFs, namely the Hawassa Lake Watershed, Sustainable Land Management Program (SLMP), New Climate and Biodiversity Conservation project, Kaffa coffee forest (NABU), Awash River Basin Soil Salinity Study (ARBSSS) project, Integrated Land Use & Development Plan (ILDP), Bale and Jamma-Urji. The MSFs focused on addressing constraints in the rehabilitation of degraded lands, Main focus Improving the environment and livelihoods of the eco-region using integrated approaches of forest management, family planning, and livelihood interventions.
Restoring, maintaining and managing the degraded land; Strengthening the capacity of the stakeholders to make informed decisions to achieve sustainable land use.  (2014)(2015)(2016). It was fully funded by the Dutch Embassy in Ethiopia and implemented by the Horn of Africa Regional Environment Center/Network unit of Addis Ababa University (HoA-REC) and ANCEDA. The project aimed to explore the benefits of the collective action approach to enable sustainable land use in Jamma-Urji.

DATA COLLECTION AND ANALYSES
Data were collected during April and August 2018 in Bale and Jamma-Urji. The Q-methodology and follow-up in-depth interviews were conducted with participants and organizers of both MSFs (see supplementary material 1). Q-methodology is a research technique combining qualitative and quantitative approaches (Watts and Stenner 2012). Q-methodology enables examining subjective viewpoints of study participants in factor analysis (Eden et al. 2005). It uses a small number of study participants to identify multiple ways of viewing a particular subject, and the ways in which those perspectives diverge Bale Eco-Region, and increase the livelihoods of local communities. The project aimed to explore the benefits of multi-sectoral and integrated approaches to enable sustainable land use in the eco-region. Accordingly, FARM Africa, SOS-Sahel, Population Health and Environment-Ethiopia Consortium, Frankfurt Zoological Society, and the International Water Management Institute established the SHARE-BER MSF in 2014.

Jamma-Urji
Jamma-Urji refers to the Jamma and Urji mountains located between Shashemene and Shallaweredas of West-Arsi zone, Oromia regional state ( Figure 1). The forest covers 6557 ha of land with altitudes ranging between 1777 and 2136 masl, and has close to 30,000 users. The site used to be covered with dense natural forests that were managed by the state until the fall of the Derg regime in 1991. Afterwards, there was no responsible organ to protect the forest, leading to an "open access" situation. The forest has been cleared for firewood and charcoal-making, resulting in severe land degradation. The majority of the population is agro-pastoralist, with an average land holding of 0.5 ha per person, and depends on natural resources for livelihoods. The main crops for consumption and sale are potatoes, maize, wheat, barley and

FIGURE 1 Location of Bale and Jamma-Urji MSFs in Ethiopia
from one another or cluster together (Lehrer and Sneegas 2018). We selected Q-methodology for this study because it helps in distinguishing viewpoints on detailed issues of communal forest governance, and enables factor analysis on the main viewpoints (Sarmiento Barletti and Larson 2019).
Communal forests were used by more than 66 PFM cooperatives in Bale and about 5000 households in Jamma-Urji. In-depth interviews were conducted with 40 informants at national and local levels in both sites (Table 2). We used purposive and snowball sampling techniques following the guidelines by Ostrander (1993). Howerver, lists of MSF participants were inadequate in both sites. Thus, we identified the organizers and participants of both MSFs for the initial interviews and then asked them to suggest MSF participants. Informants included policymakers, political leaders, agricultural experts, conservationists, CSOs, academics, farmers and village elders.
The interviews with the MSF organizers aimed to understand the theory of change through which the MSFs had been organized and implemented. Interviews with participants focused on understanding their perceptions on the equity and effectiveness of the MSF's processes and outcomes. In addition, each MSF's organizers and participants were interviewed using the Q-methodology. In this case, we examined MSF participants' perceptions on the equity and effectiveness of MSFs they had participated in. Q-method participants received 42 cards. Each card included a statement describing an aspect of an MSF -not exclusively specific to the Bale and Jamma-Urji MSFs -as participatory processes, which covered four principal characteristics: how MSFs were designed; opportunities and advantages they brought about; their challenges and problems; and alternative approaches. Participants then arranged the cards based on how much they either agreed or disagreed with each card's statement. When interviewees finished sorting out their cards, interview questions were asked to understand why they placed the cards in a specific order in order to understand their perceptions in more depth.
Follow-up interiews were also conducted with MSF particpants. In contrast to the Q-methodology, these interviews focused on the first-hand experiences of the informants in the Bale and Jamma-Urji MSFs. These interviews were used to understand the factor groups (i.e. Groups with similar opinions) within the studied MSFs. All interviews were conducted in Amharic and were recorded, transcribed and translated.
The data for Bale MSF (n=23) and Jamma-Urji MSF (n=17) were analyzed separately using the PQ Method (Schmolck 2014). Principal Component Analysis (QPCA) was conducted (see supplementary material 2 and 3). Eight unrotated factors were extracted and rotated using varimax rotation. Then, Q sorts loading significantly on to factors (>0.30) were identified. In Bale, three factor groups with eigenvalues 2 between 1.43 and 9.06 were the best fit, and explained 55% of the study variability. Three factor groups which have eigenvalues between 1.42 and 6.43 accounted for 58% of the variability in Jamma-Urji case study. While eigenvalues above 1 are often valid, factors with eigenvalue of 1.42 and above were selected because they can be explained by the statements and also align with the comments made by participants in the interviews following the Q-sorts. Some Q-sets were significantly loaded onto more than one factor. Crib worksheets were used to group statements which ranked highest and also statements which ranked lowest for each factor, and to determine the viewpoints. Data from follow-up Q-methodology interviews and field notes were used for interpretation.

RESULTS
Results indicated that MSFs played a central role in addressing constraints related to inclusiveness, equity, and transparency, which could in turn improve the governance of communal forests. Our analysis revealed a positive relationship between the perceptions on MSFs and what is needed to achieve good governance in communal forests. In particular, the type of approach employed by each MSF, coupled with the extent of enforcement of the MSF's outcomes affected their level of success in improving the governance of communal forests. Informants pointed out that using the collective action approach, which integrated capacity development, contributed to the success of the Jamma-Urji MSF, while the multisectoral approach was perceived effective in enabling joint planning and implementation in the Bale MSF.
Results showed that the approach adopted by the Bale MSF resulted in an increased level of implementation of the agreements reached by the MSF participants, compared to  (1) Participants Local (15) Government (10), Grassroots (5) Male (15) that of the Jamma-Urji MSF. In Bale, informants emphasized the success of the PFM cooperatives and multi-stakeholder taskforces in rule enforcement, such as in preventing forest encroachment and forest clearing. Both MSFs exhibited a limited representation of women and the private sector. Regarding women, the problem was more pronounced in the Jamma-Urji MSF due to the little attention given to addressing women and youth's high livelihood dependence on communal forests. By private sector we refer to individuals or companies who are 'outsiders' to local communities yet profit, or seek to profit, from community forests and their wider territory. Their lack of participation in both MSFs is notable given their engagement in the management, development, and extraction of forest resources (including timber harvesting) and tourism services in community forests.
The following subsections present the findings of the Q-sorts, which resulted in three distinct factor groups each for Bale MSF and Jamma-Urji MSF.

Factor group 1: active participation influences MSF effectiveness
Factor group 1 had an eigenvalue of 9.06 and explained 39% of the variability. The factor group indicated the belief that active stakeholder participation in decision-making and implementation processes contributes to the MSF's effectiveness. The group revealed strong agreements with statements such as "an MSF is a waste of time if its outcome is not mandatory for all relevant actors", and "for an outcome to be fair, every participant must be speaking on behalf of an interest group that selected him/her to represent them". The group strongly disagreed with statements like "an MSF's objective should be set by the convenor before including other participants" and "if participants are too transparent with information, maps, and legal documents, others may use that to further their own agendas".
13 out of 23 Q-methodology arrangements were significantly loaded to the factor and revealed that increasing an MSF's inclusiveness is more important than focusing on its outcomes. They mentioned that the mechanisms used by the MSF to achieve inclusiveness included increasing the representation of different sectors in the MSF and providing stakeholders with an equal chance to voice their concerns. The factor group underlined the limited representation of women and the private sector in zonal and wereda level MSFs, where the main decisions are taken.
Overall, the factor group supported assigning leadership positions in PFM user groups to women in order to reduce the gender participation gap. However, some informants in this factor group (> 30%) suggested that without economic empowerment and capacity building specific to women's needs, this measure would not improve women's participation. Furthermore, 2 out of 13 informants in this factor elaborated that the economic empowerment of women would make them more influential in decision-making at the community level, as opposed to assigning them seats in committees to comply with quotas and without building their leadership capacities. An informant explained the negative consequences of not using a gendered approach on an MSF's outcomes: "…one model farmer in the lowland has reached 40 bee hives. Then we [organizers] asked him "Now you are rich, what is your next plan?" he replied, "I have two wives now, I will marry a third one". We cannot blame this guy. The problem was rather ours. We would have avoided this by giving the beehive to the husband and wife and ask for their signature instead of getting only him to sign for the beehive. He would have considered the outcome as mutual and not made decisions without consulting his wife. " (Follow-up interview, MSF organizer 1, male, April 2018) Furthermore, MSF organizers underlined that the MSF would have been more effective if resources would have enabled the inclusion of more stakeholders. 4 out of the 13 informants pointed out that outcomes achieved by the MSF may not be sustained if the MSF falls short in its inclusiveness of women and private sector actors, such as investors in the agriculture and forestry sectors. Bringing in private sector actors to the MSF was perceived as one way to raise their awareness in the interrelatedness of natural resources in the eco-region, which could lead to the required behaviour change to achieve sustainable land use. At this point, it seemed difficult to assume that their interests were considered in the negotiations in their limited presence/visibility (or absence thereof) in the MSF.
Factor group 2: multi-sectoral approach enhance MSF effectiveness Factor group 2 had an eigenvalue of 2.29 and explained 10% of the variability. The factor group emphasized that collaboration among sectors in the processes increased rule enforcement and the recognition of rights. 10 out of 23 Q-method interviewees strongly favoured factor 2. The factor group strongly agreed with statements "MSFs help solve problems because they bring together government actors (e.g. development and environment planners) that would normally not work together" and "MSFs build bridges that are likely to lead to future positive outcomes (even if not right now)". The factor group strongly disagreed with statements such as "MSFs are just a way to create the appearance that participants are equals, which makes things worse for the less powerful" and "MSFs disempower grassroots organizations by giving others with less rights over their ancestral territories equal participation in decision-making".
All 10 informants in this factor group indicated that multisectoral collaboration in MSFs helped to create awareness on sustainable land use, employing integrated approaches in intervention planning and implementation, and preventing resource overexploitation in the eco-region. According to these informants, the extent of rule enforcement had increased compared to the situation prior to the establishment of the MSF due to the use of multi-sectoral approaches in MSF processes. Informants based their assessment on the periodic reports presented by MSF organizers and taskforce members in MSF meetings. An informant explained that: "The MSF comprises stakeholders from different dimensions such as academia, research, and grassroots. Such arrangement helped us to capture indigenous communities' social and economic issues. I think they would have benefited better if more stakeholders were included in the task force, as their problems will be addressed using an integrated approach. The more stakeholders, the better MSF benefits for indigenous communities will be." (Follow-up Q-method interview, MSF organizer 2, male, The informants in this factor group had positive views on the benefits of participatory approaches for achieving sustainable land use in the eco-region. They also explained that the multi-sectoral approach enabled the efficient use of financial resources in the MSF. An informant elaborated that: "Participants do their best in discussing the issues. Most of them are also committed to executing the outcomes of the forum. I think [that] the fact that concerned bodies are represented in the forum [and] discuss the problems, and suggest solutions which favour the majority, if not everyone. The mobilization and management of financial resources amongst organizers also contributes to the effectiveness of the forum. We can't even think of doing the same with the government's expenditure." (Follow-up Q-method interview, MSF participant 7, male, April 2018) However, all of the informants in this factor group agreed that although stakeholders often take assignments to implement the outcomes of the MSF, they often failed to participate in the implementation and had nothing to report on the activities in the next meetings. For that reason, an informant pointed out that measures that increased the accountability of stakeholders regarding their responsibilities could improve their performance and the effectiveness of the MSF (Followup Q-method interview, MSF participant 21, female, May 2018).

Factor group 3: addressing power imbalances increases MSF effectiveness
Factor group 3 had an eigenvalue of 1.43 and explained 6% of the variability. This factor group was explained by the attention given to addressing the negative consequences of power imbalances among stakeholders of the MSF, as shown in the strong agreement with statements such as "successful MSFs take the politics out of land use and land use change issues by making them technical" and "effective MSFs have those driving deforestation and forest degradation at the table". The factor group strongly disagreed with statements like "MSFs are often a waste of time because some participants use them to make unrelated claims" and "no matter how the is MSF designed, powerful actors always find a way to dominate the conversations held during it". 5 out of 23 Q-methodology participants were catalogued under factor 3. Informants emphasized that the intervention of PFM in establishing and strengthening PFM cooperatives and rule enforcement on behalf of task forces are mechanisms for achieving equity in the MSF's outcomes.
All 5 informants in this factor group agreed that the governance system left the zonal or regional governments with little or no power to enforce rules in the national park. This situation was a disadvantage for the local communities, who were interested in increasing their benefits from the park considering their contributions to the conservation of the park's resources (Follow-up Q-method interview, MSF participant 4, male, April 2018).
Additionally, 1 of the 5 informants in this factor group mentioned that the regulatory system has a gap in enforcing rules on investors in agriculture who process their investment licences at the federal level without the knowledge and approval of the lower-level governments. Besides, the informant elaborated that the MSF often failed to enforce rules on investors due to the lack of clarity on whether the federal or lower level governments could regulate the investments approved by the federal government (Follow-up Q-method interview, MSF participant 4, male, May 2018).
According to 2 informants in this factor group, rule enforcement on those considered as 'illegal settlers' by the local governments sometimes took political shape when the settlers from other ethnic groups or regional states such as the Southern Nations Nationalities and Peoples Regional State (SNNPR) complained to the federal government that they were excluded from the forest due to their ethnicity. The settlers were deemed illegal in conditions in which they built residences and farmed in the eco-region while not having the corresponding land use rights or land titles. An administration and security member of the multi-sectoral taskforce in Dinsho wereda also confirmed more than 2-3 incidents in the past year. The informant added that the taskforce enforced rules regardless of people's ethnicity or place of origin (i.e. whether the settlers are from within or outside Oromia).
Another informant in this factor group indicated that local communities were concerned about why the Oromia Forest and Wildlife Enterprise (OFWE) would benefit more than local communities from the forest, considering the efforts made by PFM cooperatives in conservation (Follow-up Q-method interview, MSF participant 21, female, May 2018).
Informants in factor group 3 further emphasized that the outcome of the MSF was to increase the share of forest revenues by the PFM cooperatives to 60%, while the OFWE took the remaining 40% (the proportion used to be 30% benefits for PFM cooperatives and 70% for OFWE). This outcome of the MSF has been implemented, and it enhances equity and mitigates possible conflicts among the stakeholders over benefit sharing.

Jamma-Urji
Factor group 1: ownership of processes and outcomes determines MSF effectiveness Factor group 1 had an eigenvalue of 6.43 and explains 38% of the variability. The factor group indicated that a limited ownership of processes and outcomes had constrained the MSF's success in achieving its goals. The factor group strongly agreed with statements like "an MSF is a waste of time if its outcome is not mandatory for all relevant actors" and "MSFs build bridges that are likely to lead to future positive outcomes (even if not right now)". The factor group revealed strong disagreements with statements like "an MSF's objective should be set by the convenor before including other participants" and "decision-making would be fairer if the government consulted each stakeholder group separately". 11 out of 17 Q-methodology interviewees were categorized under factor group 1. About 4 out of the 11 informants in this factor group used the phrase "The forest has no father" to express concerns on the low level of rule enforcement and sense of ownership. An informant elaborated this perspective: "We cannot bring change by saying it is not [for] me to do this, but so and so should do it. Every stakeholder needs to know that the MSF's outcomes will bring change if every stakeholder is responsible to act and also join hands to work with other stakeholders in the implementation. I really believe that implementation should be compulsory for each stakeholder, so that we move forward in the implementation." (Follow-up Q-method interview, MSF participant 1, male, July 2018) The informants in this factor group asserted that any attempt to set the agenda in the absence of the majority of stakeholders would be considered as imposition (e.g. Follow-up Q-method interviews, MSF participant 16, male, August 2018).

Factor group 2: power imbalances affect MSF effectiveness
Factor group 2 had an eigenvalue of 1.97 and explains 12% of the variability. 8 out of 17 Q-sets heavily favoured this factor group, reavealing that inequity issues among stakeholders affect the effectiveness of the MSF. All informants in this factor group stressed that the main essence of establishing an MSF is to bring the relevant stakeholders to negotiate, so that neither the government nor the NGOs could decide on behalf of all the stakeholder groups. They argued that stakeholders need to keep on negotiating how to accommodate their competing interests, explore other options together, and weigh the benefits of each option in a joint decision-making process (e.g. follow-up Q-method interviews, MSF participant 2, male, July 2018; MSF participant 12, male, August 2018).
However, they also elaborated that the government could intervenene in providing direction in cases where consensus could not be reached after extensive discussions and negotiations among stakeholders. The perspective was explained by gender inequity in the negotiation of interest among MSF stakeholders. Some informants (>50%) in this factor group recognized that the efforts to address gender inequity by allocating quotas for women in PFM committees (i.e. a group of leaders to whom the PFM cooperative members assigned leadership roles) were minimal compared to the high level of gender inequity among local communities. Informants emphasized that the situation manifested the gender norms and labour status of the households. Informants elaborated that youth and women had little opportunity to contribute to the MSF processes and outcomes, although they were highly dependent on firewood for their livelihoods. Furthermore, bias among political leaders towards their friends and relatives in the MSF also explained this perspective. 3 out of 8 informants in this factor group explained that some political leaders at kebele levels selectively shared the information on planned meetings regarding the MSF sessions with their friends and neighbours, excluding others from participating in the MSF, contributing to low levels of rule enforcement.

Factor group 3: transparency influences MSF effectiveness
This factor had an eigenvalue of 1.42 and explained 8% of the variability. The perspective was explained by attention given to bringing those aggravating deforestation and forest degradation to participate in the MSF. 6 out of 17 Q-sets heavily favoured factor 3. All informants in this factor group disagreed with the statement "successful MSFs take the politics out of land use and land use change issues by making them technical".
One informant in this factor group, a PFM committee member, stated that the MSF's lack of transparency and a proper exit strategy led to suspicion among local communities: "The MSF was working well. We (farmers) were very hopeful to sell carbon for the REDD+ program and generate income. Some experts from ANCEDA and HoA-REC were even studying and measuring carbon. We (farmers) even doubted them as they have stopped coming after taking the measurements. We (farmers) thought that some people might have already cheated us and sold our carbon. " (Follow-up Q-method interview, MSF participant 6, male, July 2018).
Informants in this factor further highlighted that the failure to provide space for stakeholders from neighboring villages to participate in the MSF negatively affected its success in bringing change. They added that natural resources are connected at the landscape level, and that local communities in the Jamma-Urji MSF should not hold discussions about the forest on their own because their decisions could affect the livelihoods of a good number of stakeholders from neigboring villages.

DISCUSSION
Our findings show that MSFs are perceived to have a positive influence on achieving good governance in communal forests. The comparative analyses reveal similarities in that participants believe that MSFs facilitate good governance by enabling participation and ownership in communal forest management. Informants seem to agree that the MSFs enabled the active participation of stakeholders in decisionmaking processes, increased trust and collaboration among stakeholders from different sectors, and supported equitable and transparent benefit sharing mechanisms among stakeholders of communal forests. The similarities could be related to the increased awareness on meeting the diverse interests of stakeholders for sustainable management of communal forests. For instance, both MSFs attempted to use local languages to increase local community engagement. The findings support the notion that stakeholder participation influences MSFs, as the level of inclusiveness of different stakeholders determines the credibility of decisions made among diverse groups of stakeholders.
Findings also reveal similarities in perspectives on the influence of power imbalances on MSF effectiveness in Bale and Jamma-Urji. Findings emphasize that power relations are important in determining the effectiveness of both MSFs. In Bale, the MSF facilitated negotiation and dialogue, which increased the share of PFM cooperatives on the revenue from the forest and wildlife. The MSF enhanced equity and reduced conflicts over benefit sharing among stakeholders. In contrast, the poor attention given to managing power imbalances constrained the effectiveness of the MSF in Jamma-Urji. Stakeholders with political powers and technical expertise in forest management dominated its sessions, leaving little room for less powerful stakeholders to participate, including women and youth. This situation could be explained by the limited consideration of socio-cultural settings in the MSF's design process.
Findings show that gender inequity has negative influence on the achievement of good governance in communal forests in both sites. However, gender inequity is more pronounced in Jamma-Urji due to the scarcity of resources to meet the growing population, and to socio-cultural constraints such as gender norms, which burden women with domestic chores. For instance, young unemployed heads of households attempted to generate income for their families by selling firewood from the communal forest. Similarly, women are forced to meet the food and energy requirements of the household by making charcoal and selling firewood from the communal forest. Our findings show that the MSFs attempted to narrow the gender gap in decision-making processes by allocating minimum quotas for women participation in PFM cooperative committees. While this measure may be encouraging, representation without empowerment might not bring the desired outcomes. Thus, the quota system per se is not adequate to address the gender inequity regarding access to and control over resources and decision-making processes in communal forests.
In fact, the practice of assigning women to PFM cooperative committees could aggravate the existing gender inequity, as putting women in time-consuming and less powerful positions could discourage them from actively engaging in decision-making processes. Women could shy away from PFM committee leadership because of their roles and responsibilities in the communities, including childcare -leaving women with little or no time for extensive meetings. In this context, women PFM committee members would be forced to support suggestions made by their male counterparts without having prior knowledge of past discussions nor having time to negotiate their interests. In the end, the committee would take men's points as conclusive, thereby raising a legitimacy question. Moreover, men tend to discuss PFM cooperative issues in informal "get-togethers", share their ideas, and lobby their interests in the absence of women PFM committee members, as Yami et al. (2013) note on the decision-making processes of communal forests in Tigray, Ethiopia. While such practice does not sound harmful by itself, it excludes women PFM committee members from accessing decision-making processes.
In Jamma-Urji, more emphasis was given to how more transparent MSFs may lead to more successful sustainable communal forest management. The different emphases could be explained by the differences in approaches in the two MSFs: diverse stakeholders were included in Bale MSF whereas only stakeholders from agriculture and natural resources departments were included in Jamma-Urji. Outcomes also differed as the Bale MSF contributed to addressing the implementation gaps associated with the non-binding nature of commitments of stakeholders, whereas the Jamma-Urji MSF failed in enabling joint planning and implementation.
The perspective that transparency influences effectiveness in the Jamma Urji MSF could be explained by the lack of clear communication strategies to enable effective communication and information exchange among the MSF organizers and stakeholders. Governments and development partners may frequently affect the level of transparency in participatory processes due to operations and agreements that are not known to the wider public. This situation could in turn affect the effectiveness of MSFs' processes and outcomes. For instance, open discussions on funding constraints could have enabled the MSF in Jamma-Urji to explore alternative mechanisms to run the MSF, such as by aligning the plans of the MSF with those of government departments.
The comparative analyses show different perspectives of an MSF's process that may enhance the effectiveness of its outcomes. Analysis of the MSF in Bale reveals that multisectoral approaches and the embeddedness of the MSF in government institutions were essential for its effectiveness. Its success is associated with strong support and commitment among stakeholders including technical experts and government agents. Our findings differ from the established literature, which tends to oppose embedding MSFs in government structures. Based on our analysis, we argue that failing to substantially engage the government in MSFs would either limit their success in bringing change to the governance of communal forests, or potentially limit these changes to the local community level. The government could play a central role in strengthening the capacities of MSFs towards improving the governance of communal forests. For instance, embedding MSFs in government structures could facilitate the alignment of efforts and mobilize the support of different government departments at low costs. However, such a move would require careful planning to avoid the government's tendency to control the decision-making process, thereby further aggravating the power imbalance.
Both MSFs in Bale and Jamma-Urji showed that adopting bottom-up approaches in designing and implementing MSFs for the improved governance of communal forests might not be practical due to the power imbalances and competing interests among stakeholders in the context of Ethiopia. Rather, MSFs would benefit from adopting a blend of bottom-up and top-down approaches. For instance, the bottom-up approach is important in mobilizing stakeholders, gathering updated evidence on challenges and opportunities, and understanding the interests of stakeholders at different levels. The top-down approach could support the implementation of MSF decisions because government authorities could use their legitimate power to guide their networks and respective institutions and mobilize implementation resources. Similarly, Khan et al. (2015) asserted that using a blended approach would maximize the effectiveness of development interventions in Ethiopia.
Furthermore, the organizers of both MSFs paid more attention to strengthening the MSFs at zonal and/or district levels than at the grassroots level, as illustrated in the multisectoral approach adopted in Bale, and in the efforts made to enhance inclusiveness and ownership of decision-making processes in Jamma-Urji. This situation could emanate from their assumption that the inputs from political leaders and technical experts at higher levels would help to make 'wellinformed' decisions. However, MSF sessions that bring stakeholders from different levels together are lacking. This finding aligns with the case in which MSFs were established to achieve sustainable forest management in Ghana and Burkina Faso, where local stakeholders' limited access to negotiation processes at regional levels influenced the effectiveness of the MSFs (Foli et al. 2018). Such disconnection between negotiation and implementation processes limits the effectiveness of MSFs.

CONCLUSIONS
This paper shows that MSFs contribute to the realization of good governance, a process in which different stakeholders engage and participate to make decisions that affect their livelihoods in an inclusive, transparent, and accountable manner as noted by Gisselquist, (2012). The case of Bale reveals that MSFs improve participation, facilitate multi-sectoral approaches, and can create opportunities to address power imbalances among stakeholders. The results are encouraging considering MSFs are 'new' forms of governance in the efforts to achieve sustainable forest management in Ethiopia. However, gender inequity, a limited sense of ownership, and a lack of transparency in the MSF constrained its success in enhancing good governance in the Jamma-Urji site. Arguably, the establishment and functioning of MSFs as projects could have limited the attention given to understanding the socio-cultural and political aspects of the stakeholders from the onset of MSFs. Aligning MSF activities with stakeholders' diverse needs and priorities and the existing norms which would favour or work against MSFs would require a much longer time than a project lifespan.
The trend in tackling deforestation and forest degradation problems in Ethiopia reveals bias towards technical solutions, such as delineating degraded land to allow for the natural regeneration of forests, construction of stone bunds, and tree planting. We believe that MSFs need to challenge such trends and attempt to understand the problems from stakeholders' realities in their everyday lives, and jointly solve these. MSFs could embrace a mix of top-down and bottom-up approaches for an increased alignment and impact to address constraints to good governance in communal forests at different levels. MSFs at grassroots levels comprise farmers and local governments, and are challenged by issues of inclusiveness and elite capture of the forums' processes and outcomes. Thus, the government and development partners should promote MSFs at grassroots level and engage in developing the capacities of stakeholders in problem identification and in the implementation of communal forest management, among others.
Furthermore, devising mechanisms to enable the economic empowerment of women and youth would reduce their dependence on forests and increase the enforcement of MSF outcomes. MSFs should invest in improving women's economic empowerment and capacity building. Contrary to the established literature, the study suggests that failing to substantially engage the government in MSFs could either limit the success of the MSFs in bringing change in the governance of communal forests or limit the changes to the local community level. The government could play a central role in strengthening the capacities of MSFs towards improving the governance of communal forests. However, such a move would need careful planning to avoid the government's tendency to control decision-making processes, thereby further aggravating power imbalances.

Instructions for Interviewees
The Center for International Forestry Research (CIFOR) is conducting a research project about Multi-Stakeholder Forums. We define MSFs as purposely organised interactive processes that bring together a range of stakeholders to participate in the dialogue and/or decision-making and/or implementation of actions seeking to address a problem they hold in common or achieve a goal for their common benefit. We would like to hold an interview with you that involves a card sorting exercise that will allow us to understand your perspective on these Forums in general. Once you have completed this exercise, I will ask you to explain to me why you have set the cards in the way you have. Please note that this discussion will be recorded because it will not be possible to take notes and discuss your selection with you. Please explain all your answers by referring to specific examples of your experience of MSFs. Your contribution is very important to us! Your responses will be anonymous. We would like to include your name in a list of people interviewed only: do you give your permission? ____ yes, that's fine ____ no, I prefer not to We will now start with the card sorting exercise.
Step 1 -Start by reading the statements in all 42 cards. As you do, separate them into three piles based on whether you Agree, Disagree, or are Neutral with each statement. You do not need to distribute the cards equally or make them match the spaces on the response grid-for now you are only going through the cards and doing a preliminary sorting. You may have noticed that all cards have phrases on one side and numbers on the other. Those numbers are random and will be used to compare how you sort the cards with how other research participants sort them Step 2 -Please sort all 42 cards onto the response grid provided, ordering them in terms of how much you agree or disagree with each statement. The bottom of the grid is numbered from -4 to 4. Please use -4 for the statements you disagree the most with, and 4 for those you agree the most with. As you will soon see, the method will force you to prioritise as you sort. We know that there might be more than two phrases that you strongly agree with (or disagree with) but, for the purpose of this exercise, you must follow the grid. You can move the cards around as much as you want until you are happy with it.
Step 3 -Once you are done sorting, I will ask you to take me through your sorting. If you feel like you want to change cards around at this point, please do so. I would like you to tell me more about the 6 statements you agreed with the most (+4 and +3), the 6 you agreed with the least (-3 and -4), and two phrases that I will select from those you sorted in the neutral area (0). For each statement, I would like you to: tell me why you set it where you did (e.g. why -3 and not -4, and vice-versa), and give me a specific example for each card based on your experience of MSFs.

Instructions for Interviewees
The Center for International Forestry Research (CIFOR) is conducting a research project about Multi-Stakeholder Forums. We define MSFs as purposely organised interactive processes that bring together a range of stakeholders to participate in the dialogue and/or decision-making and/or implementation of actions seeking to address a problem they hold in common or achieve a goal for their common benefit. We are interested in your experience of Forum X and would like to hold an interview with you during which we will ask you specific questions about your experience of it. 36. Because MSFs only address immediate problems, rather than their underlying causes, their outcomes will never change the status quo.
-1 -4 -1 37. No matter how the MSF is designed, IP/LC representatives will lack the confidence to voice their interests.
-3 -3 0 38. No matter how the is MSF designed, powerful actors always find a way to dominate the conversations held during it.
-1 -1 -2 40. MSFs disempower IP/LCs by giving others with less rights over their ancestral territories equal participation in decision-making.
-2 -1 -3 41. For an outcome to be fair, every participant must be speaking on behalf of an interest group that selected him/her to represent them.