PROTOCOL: Incentives for climate mitigation in the land use sector: a mixed-methods systematic review of the effectiveness of payment for environment services (PES) on environmental and socio-economic outcomes in low- and middle-income countries.

Abstract

Around a quarter of all anthropogenic greenhouse gas emissions originate from the agricultural, forest and other land use sector (AFOLU), driven primarily by deforestation, forest degradation and emissions from unsustainable livestock, soil and nutrient management practices (IPCC, 2014). But there is also a large potential for climate change mitigation in the sector, through removal of greenhouse gases in the atmosphere (carbon sequestration) and reduction in emissions from reduced forest and vegetation removal and improved agricultural practices. AFOLU sector1 also provides a range of other ecosystem services in addition to climate regulation. Forests and lands provide clean water, regulate soil and provide food, fuel, fiber and fresh water (MEA, 2005). Agriculture provides directly and indirectly for the livelihoods of billions of people, in addition to providing food for all the world's population (FAO, 2016a). The sector also offers livelihoods for an estimated 750 million of the world's extreme poor (FAO, ibid). Finally, forests provide paid employment for at least 100 million people and support the livelihoods of many millions more (FAO, 2016b). The United Nations Framework Convention for Climate Change (UNFCCC) have recognised the critical importance of reducing emissions from deforestation and degradation for climate mitigation (UNFCCC, 2010). In addition, the IPCC highlights the importance of preservation and restoration of other ecosystems such as peatlands and mangroves for maintaining carbon stocks and reducing emissions (FAO & IPCC, 2017; IPCC, 2014). Improved livestock and crop management also represent practices with mitigation potential (FAO & IPCC, ibid). The links between climate change, agriculture, forests and human wellbeing are complex. The world's forest area declined from 4128 million hectares of forest in 1990 to 3 999 million hectares in 2015 (FAO, 2016c). Agriculture, both commercial and subsistence, was the main driver of this global deforestation, accounting for 73 per cent of forest clearance worldwide (FAO, 2016b). This is partially driven by an increasing global demand for food from increasing incomes and growing populations, which is expected to rise 60 per cent from 2006 levels by 2050 (FAO, 2016a). At the same time, climate change is expected to negatively affect all dimensions of food security, including agricultural production of food, quality, food access through the impacts on livelihoods, and food price stability (IPCC, 2014). These complex relationships make sustainable preservation and management of forests and land, while at the same time ensuring food and livelihoods for the world's population, one of the biggest policy challenges facing the world (FAO, 2016a; FAO, 2016b). Concerns that climate change mitigation programming may have negative knock-on effects on human wellbeing and human rights, especially for the poor, remain. (Stickler 2009; Larson et al. 2013; Lawlor et al. 2013; Mutabazi et al. 2014). It is therefore important to identify strategies that reduce trade-offs between environmental protection and human wellbeing, and ideally programmes that offer win-win solutions. Economic incentives-based programmes, which aim to preserve or restore ecosystems services through financial incentives, have grown in popularity in the last two decades (Pirard, 2012; GEF, 2014; Ezzine-de-Blas et al., 2016). One such incentive-based mechanism is Payment for Environmental Services (PES). PES are a market-based approach, where users of an environmental service pay the owners or managers of the service, conditional on changes in behaviours that are likely to effect the provision of environmental services (Wunder, 2015). PES may be conditional on commitments to protect or restore forest areas or sustainable forest management, such as management of forest fires (Jayachandran et al., 2016; Alix-Garcia et al., 2014). Payments may also be tied to agricultural practices associated with reduction in GHG emissions or increase of carbon stocks, including introduction of agroforestry, silvo-pastoral or integrated crop systems, which combine crops, grazing lands and trees on agricultural land, improved tillage practices such as conservation agriculture, and reduced use of fire in rangeland management (Hedge & Bull, 2011; Garbach et al. 2012). There is some debate on the definition of PES (Wunder, 2015; Muriadian et al. 2010; Engels et al. 2008). At the most simple level, PES is a voluntary transaction between service users and service providers, conditional on agreed rules for natural resource management that aims to generate environmental services or benefits that are felt off-site, for example carbon sequestration (Wunder et al. 2015). In practice, the service “user” is typically a government or NGO acting on behalf of beneficiaries of the environmental service and the service “providers” are individuals, households or community organisations that own or manage the land or forest areas in the programme. There are a number of long-standing PES programmes in existence around the world, for example the Pago por Servicios Ambientales-Hidrologico (PSAH) in Mexico and the Sloping Land Conversion Programme (SLCP) in China. The PSAH in Mexico makes payments to landowners conditional on maintenance of certain level of forest cover, according to five-year contracts (Alix-Garcia et al., 2014). If forestland is converted to another land use such as agriculture, the landowner is removed from the programme. The SLCP in China is a large-scale programme that aims to incentivise the conversion of cropland back to forests or grassland through cash and in-kind payments to participating households, to reverse or prevent soil erosion and desertification (Démurger & Wan, 2012). In addition to these long-standing programmes, the number of new PES programmes has grown rapidly in the last decade (Börner et al., 2017). They increasingly also include goals around poverty alleviation. For example, while the original goal of the PSAH was to maintain the provision of hydrological services from Mexico's forested land, in 2006 the objectives were extended to alleviating poverty (Alix-Garcia et al., ibid). Because of the restrictions around land use from participating in the programme, implementers of PES programmes sometimes combine them with other activities to support behaviour change, such as awareness raising activities around environmental conservation or capacity building in sustainable resource use (Sharma & Pattanayak, 2015). In some cases they are also combined with more extensive support for livelihoods development. For example, a REDD+ pilot programme in Nepal made incentive-based payments to Community Forest User Groups (CFUGs). In addition to forest carbon monitoring, this programme included awareness raising and capacity building for improving local livelihoods and the use of alternative fuel and cooking technologies (Sharma & Pattanayak, ibid). Payments for Environmental Services (PES) are frequently framed as a response to “market failure” (Arriagada & Perrings, 2009). A market failure occurs when the market does not provide a socially optimum level of a service or good because of the presence of positive externalities for society from providing the service. Carbon sequestration is an example of a public good with positive externalities felt at the global level (Alix-Garcia & Wolff, 2014). While households may get some individual benefits from environmental practices such as keeping trees on land, the larger benefits are felt externally but households are not compensated financially for these external benefits by market mechanisms. This leads to household or individual decisions that are sub-optimal for society, like deforestation. The overarching theory of how PES works is quite simple. It is designed to act as an incentive for a household or community to contribute to the provision of a socially-optimal level of environmental services, thus correcting the market failure. Figure 1 presents a programme theory for how PES may influence environmental and socio-economic outcomes. The outcomes presented in the model are not the only potential outcomes of PES programmes, however we have chosen to focus on those that are of direct interest in this review. PES programme theory How PES may influence environmental outcomes The intervention aims to influence environmental outcomes primarily through provision of a positive financial incentive to change environment-related behaviours (Pattanayak et al., 2010). Cash or in- kind payments are typically made to participating individuals, households or communities on a regular basis, conditional on the environmental behaviour, for example, payments to landowners to avoid deforestation on their land. Payments may come from private actors that directly benefit from the environmental service, but more typically come from government or non-governmental organisations acting on their behalf. If a participating household or community organisation fails to uphold the minimum environmental service provision, payments are suspended. The theory underlying PES is that the financial incentives motivate participants to comply with the rules of the programme, resulting in improved land or forest management practices (Alix-Garcia & Wolff, 2014). The theory is that the increase in take-up of these improved practices will ultimately restore, maintain or enhance the provision of the environmental service that has wider benefits for society. The theory assumes that the conservation payments outweigh the benefits derived from business as usual, such as converting forests to agricultural uses, or harvesting wood for energy. PES may have positive or negative spill-over effects on land that is not enrolled in the programme. If households or communities do not enrol all their land in a programme, resource exploitation pressures may simply move on to the non-enrolled areas, known as leakage or substitution effects (Sills et al., 2008). Similarly, increased household income because of the PES programme may have implications for spending patterns and put increased pressure on local resources (Börner et al., 2017). Conversely, positive spill-overs may occur due to increased forest monitoring resulting from the program or changes in social norms relating to resource use. Such indirect effects can affect the magnitude or even the direction of the effect of a PES programme (Pattanayak et al., 2010). How PES may influence environmental and socio-economic outcomes While not originally intended as a tool for poverty alleviation, PES may increase income for complying individuals or households. To directly increase household income, the assumption is that the cash transfer is greater than lost rents previously generated from the enrolled land. Alternatively, payments may also indirectly act as an incentive for households to diversify towards other livelihood activities that are less reliant on practices that reduce the provision of the ecosystems services. For example, participants may move away from agriculture that relies on regular forest clearing towards sustainable forest activities. However, there are potential trade-offs between poverty alleviation and environmental goals. The effectiveness of PES in improving environmental outcomes is theorised to depend on effective targeting towards those actors that are the biggest threat to the provision of the environmental service (Wunder, 2007; Börner et al., 2017). If the biggest threat comes from larger, better off households or communities, the payment is best targeted towards them, but this will come at the cost of income transfers to poorer families that could support poverty alleviation (Alix-Garcia & Wolffe, 2014). A range of programme design, implementation and contextual factors may influence the effectiveness of PES programmes. Below are some key design, implementation and contextual variables that are frequently theorised to moderate the effectiveness of PES schemes. In many cases, the theory is not conclusive on whether the impact on effectiveness would be positive or negative and thus on the direction of effects of PES schemes in general (Ferraro, 2017; Pattanayak et al., 2010). These factors will be explored in the review in the analysis of heterogeneity. Targeting can influence whether PES programmes achieve their objectives. PES programmes are typically voluntary and there is a risk that households that already meet conditions will self-select into the programme. Depending on the opportunity cost of participating in the programme, households may choose to not enrol or only enrol some of their land (Ferraro, 2017). Land enrolled in PES programmes may therefore be land with the lowest value in terms of exploitation potential and thus the least likely to be exploited in the absence of PES. The result of this would be little or no added benefit of the programme in terms of environmental outcomes as land owners may have preserved resources even in the absence of payments. The lack of additionality may therefore be more prevalent where pre-programme compliance with PES conditions is high (eg: low levels of resource exploitation, as indicated by low baseline deforestation rates for example). Thus, programmes targeted to land that is at a high risk of exploitation may result in higher levels of resource protection. However, this involves predicting where landholders will exploit resources in the future, information that is generally hidden from the policy-maker implementing the PES programmes (Alix-Garcia & Wolffe, 2014). Review of existing literature There is an emerging impact evaluation literature on payments for environmental services (PES) programmes. A 3ie evidence gap map (EGM) published in 2016 identified 412 experimental or quasi-experimental evaluations of PES programmes globally, with most taking place in Low-and Middle-Income Countries (L&MICs). We are only aware of one systematic review on the effectiveness of PES, published in 2014 (Samii et al., 2014). There have also been a large number of non-systematic literature reviews, either presenting narrative discussions on the effectiveness of PES (Börner et al., 2017; Pattanayak et al., 2010; Alix-Garcia & Wolffe, 2014) or presenting a range of effect sizes for PES programmes (Ferraro, 2017). There are several reasons that warrant an update and extension of the Samii et al. (2014) systematic review. Firstly, the search for the review was completed in August 2013. 3ie's Evidence Gap Map of land use and forestry programmes (Snilstveit et al., 2016) identified at least six new evaluations of PES programmes that have been published since then, including studies from Uganda, Ecuador, Tanzania and new evaluations of long-term programmes in China, Mexico and Costa Rica. Secondly, Samii et al. (2014) were unable to do a meta-analysis for income and poverty related outcomes and for forest condition due to lack of data and heterogeneity between studies. Given the increase in the evaluation evidence base since then, we hope to be able to undertake additional meta-analyses. Thirdly, Samii et al.'s review focused on PES for forest areas. We will expand the scope of the review to include PES in other settings such as farmland, mangroves and grasslands. A number of PES programmes target other important environmental behaviours of relevance to climate change mitigation programming, for example payments to incentivise farmers to take up agroforestry on their farmland (Hedge & Bull, 2011). This will be the first review that we are aware of to systematically cover the literature on the effectiveness of PES in these areas. Finally, this review will answer new questions around design, implementation, context and costs of programmes, in addition to assessing programme effects. In doing so we will look at a broader range of literature, including process evaluations, programme documents and associated qualitative studies for the programmes evaluated in included impact evaluations. Relevance to policy and practice It is estimated that additional global investments of US$35 billion in the agriculture sector and US$21 billion in the forestry sector will be needed by 2030 to mitigate the effects of climate change (UNFCCC 2009). At the landmark United Nations Climate Change Conference (COP 21) in 2015, countries agreed to conserve and enhance sinks of greenhouse gases, including forests (UNFCCC, 2015). To ensure resources are used effectively to achieve agreed mitigation objectives it is important to ensure that decision-makers have access to reliable evidence. The United Nations Reducing Emissions from Deforestation and Forest Degradation mechanism (REDD+) is one of the main frameworks for making payments to L&MICs to preserve and sustainably manage forests. There are significant resources pledged to the REDD+ initiative. At the COP21, Germany, Norway and the UK announced that they would provide US$ 5 billion between 2015 and 2020 to forest countries if they could demonstrate verified emissions reductions (BMUB, 2015). The UN-REDD Programme currently supports 64 countries across Africa, South and East Asia and Latin America and the Caribbean to enable their participation in REDD+, and 47 so far have qualified (UN-REDD, 2016). PES are promoted as an important tool by REDD+ and are supported by a range of actors, from national governments to multi-national institutions such as IFAD, UNDP and the World Bank (GEF, 2014). The number of PES programmes operating in L&MICs has rapidly increased. A recent global review of PES identified hundreds of programmes mentioned in the literature, with 55 programmes currently in operation around the world that clearly fit the classic definition of PES (Ezzine-Blas et al., 2016). The Global Environmental Facility (GEF) alone has supported 57 projects containing elements of PES since its inception, totalling investments of over $225 million, in addition to $1.59 billion leveraged from co-financing (GEF, 2014). Despite their popularity, key policy questions around the effectiveness of PES remain unanswered (Samii et al., 2014; Ferraro, 2017; Le Velly & Dutilly, 2016). One of these questions is the extent to which the environmental and poverty reduction goals of such a programme conflict or present strategies that can generate both environmental and poverty reduction benefits. A second, and equally important question is if PES generate environmental benefits that are additional to ‘business as usual’. To meet UNFCCC emissions targets, governments implement PES programmes on the assumption that by compensating some groups to reduce their emissions, emissions in other sectors are offset (Nhantumbo & Camargo, 2015). Evaluations of PES programmes finding small effects have led some to dismiss it as an important mechanism. Indeed, a recent FAO-IPCC (2017) report on climate change and land use following the Paris Agreements stated that “[PES] effectiveness, however, is limited and they are more readily applied in some sectors (e.g. forest management) than in other emerging concerns (land restoration, soil health and soil carbon)” (FAO-IPCC, 2017: 28). The report concludes that for PES programmes to be effective, they must be better designed and informed by meta-analysis of the effects of previous programmes. A range of policy alternatives to PES exist, including private sector zero-deforestation commitments (Climate Focus, 2015) and community forestry initiatives (Agrawal & Angelsen, 2009; Angelsen, 2009). The effectiveness of many of these approaches is also contested and should be subject of future reviews. While PES may be one of the most popular policy tools in the sector, it is important to assess the relative costs and effectiveness of the approach, facilitating comparison with other options in the future. Given the resources dedicated to PES and the global importance of effective climate change mitigation activities, it is essential that rigorous and comprehensive evidence is available to policy-makers and implementers. To help inform decisions about how to use available resources most effectively we will provide a comprehensive review and synthesis of the evidence on the effects of PES, including an assessment of how intervention design, implementation and contextual factors moderate outcomes and cost-effectiveness. The objective of this review is to assess the effects of PES programmes on environmental and socio-economic outcomes in low- and middle-income countries (L&MICs). This will include identifying and synthesising evidence on how PES programme effects vary by programme design, implementation, context3; and by sub-groups of PES programme participants. We will also attempt to assess the cost-effectiveness of PES programmes. To address these objectives, we will answer the following questions: The review will follow the Campbell and Cochrane Collaborations' guidelines to systematic reviewing (The Steering Group of the Campbell Collaboration, 2016; Hammerstrøm et al., 2010; Higgins & Green, 2011; Shadish & Myers, 2004). The review will also draw on the concepts of theory-based impact evaluation (White, 2009) and theory-based systematic reviews (Snilstveit, 2012; Waddington et al., 2012) to provide a mixed-methods systematic review and analysis along the causal chain, to also address questions related to intervention design, implementation and context. To do so we will include studies in two phases. To address questions 1a, b and c, we will include studies meeting the impact evaluations study design criteria, presented below. To address questions 2, 3 and 4, studies that meet these criteria will be used as the basis for a second, targeted search to identify and include qualitative studies, project documents, process evaluations and cost data on the programmes examined. Types of population We will include studies of programmes in countries classified by the World Bank as lower income, lower-middle income, or upper middle income (L&MICs). We use the classification of the country in the year of the initiation of the program under study. There are several reasons why we decided to focus on L&MICs only. Some scoping of the literature suggests that the impact evaluation literature on PES from high-income countries (HICs) is significantly smaller and does not typically use methods that would be included in the review (Snilstveit et al., 2016; Schomers & Matzdorf, 2013). It does not typically self-identify as PES (Schomers & Matzdorf, 2013; Ezzine-de-Blas et al., 2016) and would likely result in a need to search a separate literature. This is likely to add a significant amount of work to the searching and screening with only a potentially very small number of included studies. In addition, L&MICs contain most of the world's tropical forests, which offer the greatest potential for climate change mitigation in the AFOLU sector, such as climate regulation, watershed protection and carbon sequestration (Pattanayak et al., 2010). Similarly, the findings from the HIC literature will be less relevant for mechanisms such as REDD+. Finally, given that one of our main objectives is understanding the potential for PES to offer “win-win” environmental and poverty alleviation solutions, L&MIC settings offer a more likely setting for answering this. We will include studies targeted at populations living in or near to forests, agricultural land, wetlands, grasslands and mangroves. Forests are defined as an area over 0.5 hectares with trees higher than five metres and canopy cover more than 10 per cent (FAO, 2012), including mangrove forest areas. Grasslands are areas with tree or shrub canopy cover below 10 per cent but with herbaceous plant cover (FAO, 2005). Studies of programmes in HICs will be excluded. Types of interventions We will include studies of PES programmes, defined as those providing payments to owners or managers of land, conditional on some minimum environmental/ ecosystems service provision. Payments can be either cash or in-kind material transfers, such as seedlings, api-culture and fencing. Ecosystems services are defined as the benefits that humans get from ecosystems (MEA, 2005). In ideal type PES programmes, payments are conditional on the provision of the ecosystem service itself, for example payments for increased carbon sequestration in forests (Le Velly & Dutilly, 2016). However, in practice most PES program payments are conditional on changes in behaviours that are likely to affect the provision of the ecosystem service, for example reducing deforestation or planting trees on agricultural land. We will include payments tied either to the provision of an ecosystem service or to any of the following practices related to climate-regulating ecosystems services: forest protection or regeneration; sustainable forest management practices; sustainable watershed management; sustainable agricultural practices; sustainable livestock management. The payments can be made to an individual, household, community or organisation and can either be conditional on a specified environmental commitment, for example on the fulfilment of an obligation to maintain a certain forest cover on land, or paid in advance of the PES programme. We will not limit inclusion of these programmes by the funder/ implementer (private versus public for example) or status of land (private land or state-owned/ protected land). Finally, we will include programmes that study PES alone or in combination with other intervention activities, for example interventions supporting alternative livelihoods.4 Types of outcomes We will include studies that assess the impact of PES on either environmental, socio-economic or intermediate outcomes, as defined below. PES programmes often have multiple objectives, related to both the preservation or restoration of environmental services and human welfare. There is a considerable literature on the potential trade-offs or complementarities between these objectives. By looking at both sets of outcomes, we aim to inform this debate. We will also include studies that assess intermediate outcomes such as changes in agricultural, forest or land management practices. This will allow us to report on effects at earlier stages of the PES causal chain. Intermediate outcomes We will include studies that assess changes in land or forest management practices, defined as measures of the type, frequency, intensity or adoption of such practices at the household or community level. We will also include studies that assess the adoption of sustainable agricultural practices or technologies, for example incorporating trees into agricultural or grazing lands. We will also assess measures of forest dependence, for example resource extraction. Environmental outcomes We will include environmental outcomes that are related to greenhouse gas emissions or carbon storage/ sequestration. This includes both direct measures of emissions (CO2, CH4, N20) or carbon storage/ sequestration and proxies for such outcomes. Based on previous mapping work in this area, we know that there are few evaluations that measure provision of environmental services such as carbon sequestration (Snilstveit et al. 2016). Proxy outcomes include deforestation rate, forest cover, forest condition/ degradation, forest fires, soil quality, and so on. We will accept whichever measure is used by the study authors. Once we have identified all studies, we will map all outcomes to determine if they are sufficiently similar for meta-analysis. We will also include outcomes related to the spillover effects of PES programmes on to land or forests not enrolled in PES programmes. Socio-economic outcomes We will include any measures of socio-economic outcomes, including income, consumption, well-being, livelihood security and assets of communities / households / individuals participating in PES programmes. We will also include measures of food security across the four dimensions of food availability, access, utilisation and stability included in the Declaration on Food Security (FAO 2009). These include food consumption, food expenditure, prevalence of undernourishment and nutritional status (FAO 2013). We will accept whichever socio-economic measure is used by the study authors. Once we have identified all studies, we will map all outcomes to determine if they are sufficiently similar for meta-analysis. We will include studies in two stages, in a similar approach to Snilstveit et al. (2015). In the first stage, we will include studies that assessed the effects of interventions using experimental designs or quasi-experimental designs with non-random assignment that allow for causal inference (to address primary research question 1). Specifically we will include the following: Ferraro and Miranda (2014; 2017) argue that combining panel data with baseline observations and statistical matching is the most effective quasi-experimental method at reducing bias when evaluating conservation sector programmes. However, given the expected small size of the evidence base, we will include studies with post-intervention outcome data only as long as they use some method to control for selection bias and confounding. To account for the differences in the quality of study designs and analysis methods, we will appraise the risk of bias in all included studies and do sub-group analysis by risk of bias status. Before-after studies and observational studies without control for selection bias and confounding will be excluded. Additionally, modelling based studies, commentaries and literature reviews will be excluded. To address questions 2 and 3 on programme design, implementation and context, we will extract descriptive and qualitative data from the included experimental and quasi-experimental studies. In addition, we will conduct a targeted search for additional papers on the programmes