Can multilevel governance transform business- as-usual trajectories driving deforestation? Lessons for REDD+ and beyond

• Coordination across levels and sectors cannot always be achieved through negotiation: REDD+ and similar initiatives must go beyond technical criteria, engage with politics and support social movements to strengthen transformative coalitions. Anne M. Larson and Antoine Libert Amico, with Christopher Martius, Ashwin Ravikumar, Jazmín Gonzales Tovar, Laura Kowler, Rodd Myers, Dawn Rodríguez-Ward, Anna Sanders, Tim Trench, Anastasia L. Yang, and Paulina R. Deschamps DOI: 10.17528/cifor/007043 | cifor.org


Introduction
Reducing emissions from deforestation and forest degradation and enhancing forest carbon stocks (REDD+) is an inherently multilevel process that requires collaboration across diverse levels and sectors of governance. As with other global initiatives, however, implementation of REDD+ on the ground has caused friction with actors at various levels (Sanders et al. 2017;Myers et al. 2018;Rodríguez-Ward et al. 2018).
To address these concerns, CIFOR undertook research on multilevel governance and carbon management at the landscape scale. Researchers asked: How are land-use decisions made across multiple levels and sectors? How is REDD+, as an idea, or as a specific project or initiative, interacting with these political realities? And what lessons can be learned that might help improve equity and effectiveness to shift the current trajectories of land-use change? Emerson et al. (2012) define multilevel governance broadly as the "processes and structures of public policy, decision-making and management that engage people constructively across the boundaries of public agencies, levels of government, and/ or the public, private and civic spheres in order to carry out a public purpose that could not otherwise be accomplished." The concept thus refers to how government institutions can negotiate power relations among levels, sectors and actors (Myers et al. 2018). That is, for the purposes of this brief, governance refers to how government gets things done across public, private and civic spheres, with the practice of governance at least as important as its structures.
In this brief, we examine how forest and land-use policies are translated top-down through multiple levels (Sanders et al. 2017;Trench and Libert Amico, in press) and how actors on the ground communicate upward (Rodríguez-Ward et al. 2018). After presenting our research methods, we examine multilevel governance challenges related to deforestation drivers, tensions among different levels of government, responses of local actors to top-down projects and multi-stakeholder coordination.

Methods
As part of CIFOR's ongoing Global Comparative Study on REDD+, this research project aimed to improve our understanding of how land-use policies are defined and interpreted in practice. This brief refers primarily to results from fieldwork between 2013 and 2017 in five countries: Indonesia, Mexico, Peru, Tanzania and Vietnam. It synthesizes published research results to highlight main discussions and findings.
We selected two subnational jurisdictions (e.g. province or state) 1 in each country, prioritizing regions with REDD+ initiatives. In each jurisdiction, we selected approximately five landscapes as case studies (for a total of 54 study sites). We conducted over 700 semi-structured interviews with actors including local inhabitants, local and national governments, nongovernmental organizations (NGOs), indigenous and community organizations, and private companies. Table 1 provides an overview. 2 We also produced several complementary studies. In each country, legal reviews examined the roles and responsibilities of different levels of government in land use and landuse change. 3 Furthermore, participatory multi-stakeholder workshops developed divergent future scenarios of land use in eight landscapes in Indonesia, Mexico, Peru and Tanzania (Ravikumar et al. 2014(Ravikumar et al. , 2017a. 4 In Mexico and Peru, we also studied monitoring, reporting and verification (MRV) systems for REDD+ through semi-structured interviews with government and civil society organizations in Mexico (49 respondents) and Peru (32 respondents). 5  (Yang et al. 2016). A multilevel governance approach evokes attention to these multi-scalar drivers of deforestation and degradation.
Evidence suggests that REDD+ initiatives do not address the key underlying drivers of land-use change (Salvini et al. 2014;Duchelle et al. 2018). Ironically, REDD+ strategies have failed to see the forest for the trees. They often focus on minor, direct drivers of ecosystem degradation (for example, by smallholders) rather than on interests and incentives that promote a businessas-usual agenda of forest-destructive practices.
In Peru, for example, government reports identify 'migratory agriculture' as the main driver of deforestation in the Amazon. Although smallholders are directly responsible for some deforestation, the analysis -based on satellite remote sensing of patch sizes -conflates various drivers, including migrants to the forest frontier, shifting cultivation and a variety of both sustainable and unsustainable agricultural production practices (Ravikumar et al. 2017b). Underlying drivers of such behavior, including agricultural credit schemes and subsidies, are overlooked Ravikumar et al. 2017b). Moreover, these official reports divert attention from businesses that profit from and promote deforestation.
In Tanzania, REDD+ initiatives often focused on land degradation caused by the poor (e.g. small-scale charcoal producers) rather than at what motivates these activities (e.g. urban charcoal demand) (Kijazi et al. 2017). Findings from Vietnam indicate that conversion of forests to rubber plantations negatively impacted smallholder livelihoods by causing food insecurity, leading smallholders to clear other forests to cultivate subsistence crops (Yang et al. 2016). The focus on smallholders and local livelihoods fails to address key drivers such as rubber and hydropower companies, as well as underlying issues of land tenure (Yang et al. 2016). Similarly, in Indonesia, drivers of land-use change such as oil palm plantations are closely tied to weak land tenure security and poor enforcement of social and environmental protection laws (Myers et al. 2016).

Tensions among levels of government
Subnational governments have assumed important landuse roles ranging from specific attributions (e.g. the granting of permits, concessions and land titles) to participation in policy design. The countries studied have gone through distinct decentralization processes. In interviews, government representatives often referred to limitations in the scope and/or results of these processes, confirming earlier findings, "democratic decentralisation, even where legislated, is rarely implemented well" (Larson and Ribot 2009, 176;Rodríguez-Ward et al. 2018).
In the Socialist Republic of Vietnam, the central government is present at all levels of subnational government through sectoral ministries. The Republic of Tanzania is similar, in that the Prime Minister's Office -Regional Administration and Local Government has its own regional and local dependencies; Zanzibar, for its part, has an autonomous central authority.
In Mexico's federal system, states can transfer powers to municipalities within their territory except those related to natural resources, which fall under federal control. In the presidential constitutional democracies of Indonesia and Peru, some specific attributions have been devolved to each level of subnational government. However, the higher-level subnational governments (provinces in Indonesia and regions in Peru) retain most of the powers transferred from the central government (though this has gone back and forth over time in Indonesia). Furthermore, national governments have strategically retained some key legal powers. In Peru, these powers include strict oversight of annual budgets for subnational jurisdictions and authority to approve large-scale mining concessions (Wieland Fernandini and Sousa 2015).
Overall, our research did not find a reversal of decentralization trends but rather tensions over roles and responsibilities. In Mexico, Libert Amico and Trench (2016) found the central government was reluctant to give up control of decision making, especially when budgets were tight. In their study of MRV in Mexico, Deschamps and Larson (2017) found subnational governments were frustrated by the difficulty of expanding the scope of the national MRV system, establishing complementarity with other national and subnational monitoring initiatives, and increasing subnational stakeholder participation. Similar challenges were found in Peru, where tensions emerged after one subnational MRV system developed faster and with greater technical detail only to be overruled by a lower-resolution national system .
The comparison between the mainland sites in Tanzania and Zanzibar demonstrates some of these multilevel challenges. In the former, REDD+ NGOs engaged mostly with district authorities; in the latter, with central authorities.
In both cases, there was greater ownership and satisfaction among the authorities who were directly engaged. In contrast, the authority that was less engaged (the national government in Tanzania and the districts in Zanzibar) reported frustration, lack of ownership and tension with the primary authority (Kijazi et al. 2017).
Another multilevel, multisectoral challenge is the failure of government to take responsibility for finding solutions to problems it exacerbates. For example, in the Peruvian region of Madre de Dios, the central government issues mining permits. However, it does little to help the regional government address the massive influx of people, illegal mining and associated pollution. It has also failed to resolve overlapping land rights covering over 1 million hectares (about 20% of the region) that emerged before titling and concession-granting authority was decentralized (Rodríguez-Ward et al. 2018). Similarly, community demands over customary land rights in response to a governmentdecreed protected area studied in Indonesia were met with different government offices simply "passing the buck" to other levels (Myers and Muhajir 2015).
An unexpected collateral effect of the distribution of legal powers is the opportunity for 'forum shopping'. Companies or other powerful economic actors can often find allies at some government level or sector who will help facilitate their interests when another level will not. For example, in one Peruvian region, when the regional government failed to respond to an oil palm company's interests, the company convinced the national Ministry of Agriculture to reclassify land in a way that allowed it to establish a plantation, despite local opposition Ravikumar et al. in press). However, decentralization can also benefit innovation, challenging business-as-usual. In one district studied in Indonesia, the district head (bupati) limited oil palm expansion and worked with NGOs to develop alternative livelihood opportunities (Myers et al. 2016).
Thus, multilevel governance has challenges, but also provides opportunities. Even in centralized regimes such as that of Vietnam, the multilevel structures gave space for local governments. However, some local authorities had greater understanding of, or concern for, local impacts and local needs than others (Yang et al. 2016).
Most governments have contradictory policies on land and natural resources, supporting exploitation on the one hand and conservation on the other. Even when the only authority granted is over implementation, local governments sometimes have leeway to decide which will be the local priority (e.g. Boyd et al. 2018, Stickler et al. in press).

Hitting the ground: Views from the bottom up
REDD+ initiatives are rolled out in geographic locations with specific histories. While this may seem obvious, the tendency to turn to top-down blueprints is striking. It suggests a lack of awareness, resistance to flexibility or a limited ability to adapt to local contexts and needs.
Top-down approaches hinder project success by impeding local actors from negotiating desired outcomes. For example, tensions emerged in Central Kalimantan, Indonesia's official REDD+ pilot province, when top-down approaches led local actors to see themselves as subjects of a lab experiment ("guinea pigs") rather than active players (Sanders et al. 2017). Local people demanded greater flexibility in project design and timeframes to provide space for learning and contributing local insights.
Unable to shake off the inheritance of other 'aid' and 'development' projects conceived in the global North, topdown approaches frequently put forth by REDD+ proponents are received with suspicion and protest. Some technical problems can be addressed through knowledge exchange, for example. However, research shows business-as-usual interests can use technical discussions to sideline other issues. Myers et al. (2018: 2) illustrate how some REDD+ projects proved to ignore political realities, turning to "apolitical and inoffensive terminologies and logics." For example, 'benefit sharing' can be framed as a 'technical approach' 6 that masks underlying political realities, thus failing to address more fundamental issues of rights and justice for forest-dwelling communities (Myers et al. 2018). Recognizing indigenous peoples and local communities as substantive rights-holders (rather than 'project beneficiaries') can help place them at the center of forest and climate initiatives (Sarmiento Barletti and Larson 2017).
While oil palm plantations in Indonesia often generate tense community relations, one company in the Koyong Utara and Ketapang districts provided substantial community benefits (Myers et al. 2016). It promoted community consultation and meaningful participation, exceeding both government and industry 7 requirements. At the same time, customary (adat) leaders facilitated negotiations and land allocation in what respondents reported as a mutually beneficial agreement, considering previous land degradation.
Implementing the 'recipe' of free, prior and informed consent can itself deny the historically rooted disadvantages of local peoples: when people's options are drastically limited by economic or other marginalization, it is unclear where consent can truly be 'free' (Sanders et al. n.d.). Projects fail to meet their intended goals because they encounter a complex reality on the ground that is poorly understood. Illegitimate processes can lead to the failure of REDD+ projects (Myers et al. 2018). In Vietnam, research showed that projects in which decision making occurred at the local level had the strongest overall procedural equity (Yang et al. 2016).
Demands for accountability run through and across all levels and sectors. For example, subnational governments and NGOs can fall into the trap of only being accountable to those 'above' (e.g. central government and funders) and not to those 'below' (also called 'beneficiaries') (Myers and Muhajir 2015;Sanders et al. 2017;Trench et al. 2018).
The analysis of stakeholder networks points to the key role of "actors in the middle" that interpret policy from the top-down and vice versa, bridging between government agencies and communities (Sanders et al. 2017;Rodríguez-Ward et al. 2018).
In Mexico, where forest-dwelling communities rely on the central government for everything from permits to subsidies, the forestry technician emerges as a crucial broker, as well as an expression of the privatization of technical assistance in the forestry sector (Trench et al. 2018). Sometimes brokers can define public investments and regional planning. They may have their own interpretations and goals, at times facilitating low emissions alternatives with more commitment than national governments Trench et al. 2018; Libert Amico and Trench in press).

Multi-stakeholder coordination as solution?
Coordination problems across levels and sectors are widespread. They include power asymmetries, barriers to information sharing , lack of clear responsibilities and of sound channels of communication (Deschamps and Larson 2017), and the failure to integrate local needs (Sanders et al. 2017). Horizontal cross-sectoral challenges at the national level -identified as being among the central challenges to REDD+ ) -persist at the subnational level (Ravikumar et al. 2015). For example, crosssectoral coordination in Mexico was reported to depend on ties between individuals in different agencies rather than on institutional practices or culture (Libert Amico and Trench 2016;Trench et al. 2018).
"Lack of coordination" is often cited in interviews as an underlying problem in promoting sustainable land use. Where interests are already aligned, clarifying responsibilities, improving information flows and ensuring a clear government mandate for coordination can support solutions (Larson et al. in press). Nevertheless, research suggests that coordination failures are often a symptom of deeper challenges based on differences of interest, knowledge and power. Hence, it is not a question of coordination per se, but of who is coordinating their efforts with whom and to what end. Actors have divergent and at times irreconcilable objectives, and political coalitions may actively undermine coalitions for sustainability and local peoples' rights. At the same time, successful coordination is driven by political organizing over time by activists, local people, NGOs and key government supporters (Ravikumar et al. in press).

Conclusions
Analyzing multilevel governance points to the complexity of issues at hand when promoting lowemissions development strategies. It also illustrates the ways in which this complexity has been used to overlook questions of inequity or to favor specific interests. Policy makers and implementers supporting REDD+ need to explicitly recognize and address the political dimensions of land-use governance (Myers et al. 2018;Rodríguez-Ward et al. 2018). They must pay greater attention to issues of power and authority around land-use rights and the underlying incentives for forest conversion (Rodríguez-Ward et al. 2018).
Recognizing indigenous peoples and local communities as substantive rights-holders can help place them at the center of forest and climate initiatives (Sarmiento Barletti and Larson 2017). Similarly, subnational governments can be a critical ally in advancing sustainable land use: even if their formal authority is limited, they can shape project implementation at the local level and develop strategic cross-country networks that may sow the seeds of transformational change (Boyd et al. 2018;Stickler et al. in press). An essential pathway toward solutions may lie in finding and encouraging supporters of sustainable land-use outcomes within government sectors that traditionally drive deforestation and forest degradation (Ravikumar et al. in press).
To succeed, REDD+ and similar initiatives must go beyond technical criteria, engage with politics and support social movements to strengthen transformative coalitions (Myers et al. 2018;Ravikumar et al. in press).