Payment for Environmental Services: Hypotheses and Evidence

Abstract

The use of payment for environmental services (PES) is not a new type of contract, but PES programs have become more in vogue because of the potential for sequestering carbon by paying to prevent deforestation and degradation of forestlands. We provide a framework utilizing transaction costs to hypothesize which services are more likely to be provided effectively. We then interpret the literature on PES programs to see the extent to which transaction costs vary as predicted across the type of service and to assess the performance of PES programs. As predicted, we find that transaction costs are the least for club goods like water and greatest for pure public goods like carbon reduction. Actual performance is difficult to measure and varies across the examples. More work and experimentation are needed to gain a better outlook on what elements support effective delivery of environmental services.

1. Introduction

There is a growing recognition that natural resources provide flows of services that in turn provide economic value. Common environmental services (ES) that benefit societies include carbon sequestration, watershed services, biodiversity maintenance, and landscape beauty. Like other public goods, these services are often underprovided, as most of the benefits are externalities.¹ Governments could address these issues through regulation or tax/subsidy programs, but in the developing world, enforcement is often both costly and difficult. Moreover, the enforcement of noncompliant poor populations is politically challenging. Alternatively, payment for environmental services (PES) is a policy tool championed to provide market incentives for the private upkeep of natural resources providing downstream users with ES. PES is defined as a voluntary agreement between a buyer and seller in which payment is given conditional on the ES being adequately provided (Wunder 2006). Many researchers and policy makers advocate the application of PES schemes to develop new market-based tools for combatting deforestation and encouraging forest conservation in the developing world over alternate policy tools.² In the absence of these mechanisms, land users will lean toward commercial use of land, often involving deforestation for agricultural purposes, resulting in depletion of the natural resource, as they have no way to monetarily realize the gains of providing the nonmarketable services (Kroeger & Casey 2007). The market-type exchange looks appealing because (a) land use changes in developing countries tend to be primarily a function of markets, not of regulations (Alston & Andersson 2011), and (b) it is seen as a way to compensate rather than regulate the poor.³ PES draws on the Coasian principle that with clear property rights and low transaction costs, bargaining can achieve the socially optimal level (Coase 1960). Often overlooked is that Coase also indicates that when transaction costs are large and negotiation is prohibitively costly, government regulation may be the best alternative.⁴

The purpose of our article is to first address the potential transaction costs for PES programs across different ES. Following the development of hypotheses, we review the extant evidence in the literature on PES programs. The ultimate question is: To what extent do PES programs work? That is, do they succeed in providing the ES, or should other policies be pursued? At issue are the transaction costs. For many services targeted by PES programs, the transaction costs may be high. The need for monitoring from an established baseline significantly increases transaction costs. In the case of ES, property rights are frequently unclear, especially in developing countries. Drawing on prior literature and our own insights, we discuss situations that can be expected to drive up transaction costs. The variables considered include both physical attributes and institutional settings. Interactions are important, and there is not likely a one-size-fits-all solution (Agrawal et al. 2011). As a result, we advocate adopting an analytical framework, such as the social-ecological-system (SES) framework, which supports comparative research of PES (Ostrom 2009). Given this, it is important to assess the extent to which PES schemes can achieve their goal of additional conservation, paying careful attention to the interaction with the contextual setting to discover what situations work best. The ex ante expectations are that transaction costs are largest in the case of carbon sequestration, lower for biodiversity service, and lowest for watershed services.

Experimentation with PES programs has begun in the real world in a variety of settings. Empirical analysis on their effectiveness remains somewhat sparse, but we review 18 studies to assess the performance of PES.⁵ The literature reviewed also provides a menu of techniques that are employed to address issues of measuring service provided, with particular attention to establishing additionality, i.e., how much additional ES is provided relative to the counterfactual of no program. Overall, we find mixed results, although the evidence supports our expectations of particular difficulties for carbon sequestration, including high transaction costs.

Finally, we assess our results in the context of the much touted initiative Reduced Emissions from Deforestation and Forest Degradation (REDD). This initiative seeks to address carbon emissions due to the loss of carbon sinks, particularly in developing countries that contain a large portion of the world's tropical forests. Deforestation is estimated to contribute nearly 20% of global carbon emissions (Palmer 2011). REDD has been expanded to REDD+ to include conservation and sustainable management to enhance forest carbon stocks.⁶ PES is currently a leading candidate for the implementation of REDD+. Incentives are seen favorably in this context due to the international nature of the issue and to the inability to regulate sovereign nations. However, transaction costs are already high in carbon sequestration, and the international setting, especially that involving developing countries, drives transaction costs even higher (Alston & Andersson 2011). It remains unclear whether REDD+ can effectively sequester additional carbon for any meaningful time period.

2. Transaction Costs and Property Rights

Coase (1960) highlights the important role that transaction costs have in dealing with social costs. Coase provides two key insights: the reciprocal nature of social costs and the importance of considering transaction costs when rights are assigned, as such costs may impede otherwise beneficial transactions. In the context of PES, transaction costs include negotiation costs, which are often exacerbated by the lack of clear property rights; monitoring (verification) costs; and enforcement costs. Ultimately, the size of the transaction costs weighs heavily on the potential success of PES schemes. When the transaction costs are large, they may completely outweigh the private gains from a trade. In such cases, PES is not an efficient mechanism to achieve conservation goals. Here we assess ex ante expectations as to which scenarios present lower transaction costs. We consider the physical attributes of various ES as well as the institutional settings. There is certainly some interplay between the two, and specific institutions may be effective in reducing transaction costs for one service but not for other services.

2.1. Physical Attributes

In practice, most PES schemes target at least one of four ES: (a) carbon sequestration, (b) watershed services, (c) biodiversity, and/or (d) scenic beauty. Initiatives targeting the first three are now somewhat common, whereas the fourth is rarely the primary goal.⁷ The physical characteristics of the service vary across these categories and impact the expected transaction costs.

These ES have attributes of a public good, but to a varying degree. The scale of the public good has implications as to the number of people involved and the extent of the free-riding problem that accompanies it. Carbon sequestration results in reduced global warming. The reduction in atmospheric carbon concentrations is one of the few examples of a pure public good. Reduced carbon is nonrival, as all those on Earth can consume the same unit of reduction. Not only can everyone consume the reduction, but everyone is forced to, as it is nonexcludable by nature. In this regard, one would expect the transaction costs to be large, as the number of parties that need to negotiate is quite large, providing ample free-riding opportunities. In addition, nation states have very different preferences as to climate change. The unproductive Doha climate talks in 2012 illustrate the difficulties of negotiating these global issues. In contrast, watershed services have been labeled a club good rather than as a true public good (Wunder et al. 2008). A club good is nonrival and nonexcludable, but on a smaller scale. In the case of watershed services, access to the service is limited to those downstream (physically) within the same water basin. In this regard, it becomes simpler to identify the beneficiaries of improved watershed services and those that should be the providers of the service (Turpie et al. 2008, Van Hecken et al. 2012). With fewer possible providers and fewer identified beneficiaries on both sides of the contract, transaction costs are expected to be lower and potential free riding to be smaller than in the case of carbon sequestration. Biodiversity services run the spectrum. In the pure case, it is a true public good like carbon sequestration in the sense that the potential gains from maintaining biodiversity can be consumed by all. However, in most cases in which PES targets wildlife preservation, ecotourism is a central part of local economies. In these cases, a private benefit derived from the broader service provides a marketable and profitable good.⁸ In these instances, a beneficiary is easily identifiable, resulting in decreases in transaction costs and possible free riding. In contrast, in areas with no ecotourism, biodiversity services will suffer the same free-riding issues as carbon sequestration.

In addition to the variation in demand, supply of ES also differs. Location and scale of the land providing the service are integral to successful provision. Carbon sequestration exhibits a nearlinear relationship in production with regard to the land enrolled. By planting one additional tree or, alternatively, not cutting one down in a young and growing forest, additional carbon sequestration is provided. This allows almost any plot of land to provide the service, and one can target large plots of lands to reduce the relative transaction costs.⁹ Watershed services are considerably less fungible: Location matters (Kroeger & Casey 2007). The potential plots of land to provide the service are those upstream in the basin. This situation can reduce transaction costs by identifying a limited set of providers. However, watershed services do not exhibit a linear relationship with land enrolled (Swallow & Meinzen-Dick 2009). The service provision is dependent on the activities on adjacent lands that are also in the upper water basin. Indeed, successful service may require that all the land be enrolled, making it necessary to contract with all upstream parties and thus increasing the transaction costs. Biodiversity is also location dependent and nonlinear in production (Nelson et al. 2010). Often, scale is important, and migration corridors connecting the landscape are crucial to the survival of wildlife. In this regard, biodiversity services are similar to watershed services, although the benefits often exhibit attributes of a public good rather than those of a club good.

A crucial element of an effective PES scheme is conditionality. The application of this principle requires monitoring and quantifying the service provided as well as enforcing sanctions for noncompliance. Naturally, monitoring and enforcing present any provision with additional transaction costs and are not unique to PES compared with other policy mechanisms. Measuring the flow of the ES is difficult, although the difficulty is easing thanks to improving scientific knowledge and technology. In practice, most services require some land management alterations, and conditionality is often employed at this level. Although land use is easier to observe and measure than other ES, even land use is subject to the question of additionality: whether payments induce additional inputs toward producing the service. We discuss below the use of the correct counterfactual or business-as-usual scenario.

Watershed services and wildlife preservation provide a more feasible setting for relatively low cost monitoring of output. Watershed services encompass quality, quantity, and reliability of flow (Kosoy et al. 2007). All three of these are possible to test and measure in any given watershed by gathering both before and after data. Additional econometric controls may be necessary to infer causality. Turpie et al. (2008) note that after the implementation of the Working for Water (WFW) program, water flow increased 46 million cubic meters a year. Likewise, in situations aimed at augmenting species populations, sample counts could be taken and some idea of improvement in habitat quantified, as was done for a program in Cambodia (Clements et al. 2010). In contrast, measuring carbon sequestration attributable to a particular plot of land is less direct. The transaction costs are likely prohibitively expensive to truly quantify the service provided. In this regard, carbon sequestration programs will likely employ conditionality on land use and scientific estimates of sequestration based on land coverage.

2.2. Institutional Attributes

PES schemes may be pursued in various institutional settings and vary in their institutional structure. In regard to design, the source of funding is an important distinction among PES schemes (Corbera et al. 2009, Engel et al. 2008). Broadly, the source of funding is either from the user or through the government. User-financed programs are expected to have relatively larger transaction costs than government-financed programs because of economies of scale in transaction costs. A large portion of the transaction costs are in start-up costs (Börner et al. 2010, Hegde & Bull 2011, Wunder & Albán 2008). Given that government programs are typically larger than private programs, they enjoy economies of scale in reducing transaction costs per service provided (Blackman & Woodward 2010, Wunder et al. 2008). Allowing the government to negotiate reduces the number of parties directly involved while controlling the free-rider issue through the power of taxation or regulation. User-financed programs struggle to initiate due to large start-up costs. Although government-funded programs tend to have lower negotiation costs, there is a concern that other transaction costs—in particular monitoring costs and at times enforcement costs—may be higher. User-financed programs value the service enough to take on costly transactions to receive the service, suggesting an incentive to monitor the service and to ensure that it is provided and enforced (Corbera et al. 2009). Vatn (2010), along the lines of Coase (1960), points out that the greater efficiency of user-financed programs is only an “efficiency net” of the transaction costs. In Coasian terms, all transaction costs need to be fully considered in assessing social costs. Given the economies of scale in transaction costs, classifying user-financed programs as more efficient, as suggested by some scholars, may not be entirely accurate.

2.2.1. Property rights

The assignment of property rights is at the core of Coase (1960). In the case of PES, the property rights over the externality become clearly defined in that the potential provider has the right not to provide the ES.¹⁰ What remains less clear is who owns the rights to manage the land from which the ES flows, particularly in developing countries. Often, there is a split between de jure property rights and de facto property rights. In many situations, the land is government owned or private land is regulated, but these de jure statutes are not enforced (Galudra et al. 2011). In addition, locally enforced rights are often without judicial backing by formal legislation (Alston & Andersson 2011). The issue is critical because as much as 2/3 of future deforestation is predicted to occur where tenure is ill defined (Börner et al. 2010). Given the government's inability to enforce de jure rights over de facto rights, there arises an incentive to have user-based PES plans that include all the owners and users of the resource (Muñoz-Piña et al. 2008).

Many PES programs have been implemented whereby the activity is already required by law but poorly enforced (Börner et al. 2010, Clements et al. 2010, Honey-Roses et al. 2011, Turpie et al. 2008, Wunder & Albán 2008). Property rights must be established prior to the PES program, although some scholars submit that we need only rights over the service flow (or de facto use rights), and not the actual land ownership (Lyster 2011). For instance, in the context of carbon sequestration, there is consideration of carbon rights, creating an asset of the carbon sequestered in the vegetation of the land separate from the land itself. This framework provides a mechanism to give indigenous forest users access to the PES market for activities on state-owned forest. Karsenty et al. (2013) provide a good discussion of this issue and overall find it lacking because the concept is poorly defined. The authors note that providing use rights separately may provide leverage to future land tenure reform but may also crowd out the local users. They reaffirm other scholars' position that having secure rights of exclusion and management is sufficient to successfully provide the service and engage in projects (Mahanty et al. 2013, Palmer 2011, Swallow & Meinzen-Dick 2009, Wunder 2006). For instance, individuals renting land compose 1/3 of the participants of a PES scheme in the United Kingdom (Dobbs & Pretty 2008). Engel & Palmer (2008) provide an example in Indonesia in which communities with unclear tenure are able to exclude logging operations and to effectively issue logging permits. That said, security of tenure does exert the largest influence on the likelihood that a provider will opt into a PES program (Zbinden & Lee 2005). Furthermore, the security of the tenure may not be exogenous, as the presence of a PES program can add value to the land and provide the incentive to establish property rights (Alston et al. 2012, Engel & Palmer 2008).

Whereas clear property rights are necessary, private property rights are not. Communal property rights offer economies of scale in transaction costs much as government-financed programs do. Transaction costs can be reduced by enrolling large sections of lands rather than many individual smaller plots. In this respect, communal land reduces transaction costs (Akiefnawati et al. 2010, García-Amado et al. 2011, Swallow & Meinzen-Dick 2009). The effectiveness of such deals depends on the ability of the community to engage in cooperation and to ensure delivery of the service. To this end, PES activities are likely to function better in communities that have proven success in cooperation or that at least exhibit attributes associated with successful resource governance (Mahanty et al. 2013, Ostrom 2009).

2.2.2. Monitoring and leakage

Issues with monitoring and actual performance are one of the largest hurdles in a PES program. Engel et al. (2008) highlight many of the issues. For one, it is difficult to measure the service itself. An alternative, used by many programs, is to make payments conditional on inputs rather than on outputs. Focusing on a change in land use is easier to observe than measuring the elusive service that it provides. Notably, technology is making this process more transparent, reducing monitoring costs in the process. With the increased availability of satellite imaging, it is possible to remotely assess changes in land usage. Even with the increasing ease afforded by monitoring inputs, it remains difficult to establish a credible counterfactual. That is, how much additionality is achieved through the payments versus what would have occurred in the absence of the program? In a market-type situation, additionality is important, or else payments merely transfer wealth, with no additional service provided.¹¹ Additionality is easier to identify and monitor in programs that focus on changing current land use rather than on payments to avoid possible future changes such as deforestation, lowering the transaction costs (Andersson & Richards 2001). The lack of low-cost monitoring technology presents a large hurdle to the effectiveness of any market-based system of ecosystem provision (Kroeger & Casey 2007). Although monitoring inputs provides a cheaper option aided by technological advances (Stickler et al. 2009), it does not ensure that the service purchased is actually provided, presenting a possible trade-off in transaction costs and effectiveness.

Programs may be more effective when they are larger in scale by minimizing leakage (Busch et al. 2012, Sierra & Russman 2006, Swallow & Meinzen-Dick 2009). Leakage in small PES programs may occur for two reasons. First, individual land owners may enroll part of their land in a program and then produce private goods on an alternative plot of land that otherwise would have been provided by the ES. Potential leakage is smaller for watershed programs for which only specific plots of land impact the service. Programs can address this issue by the incorporation of additional clauses when the contract is negotiated. A higher level of specificity in the contract may raise transaction costs in negotiation, but contracting over larger plots of land reduces transaction costs per hectare, as discussed above. The second type of leakage is due to market influence. If supply of a good is reduced, market prices increase and induce other independent land owners to bring marginal land into production. For instance, the Conservation Reserve Program (CRP) (organized by the US Department of Agriculture), which pays farmers to take land out of production, is estimated to suffer 20% leakage stemming from higher commodity prices from reduced supply (Claassen et al. 2008). A closer screening of the exact nature of the transaction costs of all parties entailed may reduce this leakage. Alternatively, as Coase (1960, p. 39) notes, sometimes “it costs too much to put the matter right.”

2.2.3. Poverty reduction

Many programs, especially government-financed programs, aim to alleviate poverty in addition to providing an ES. When the effectiveness of a program in producing the ES is considered, the pro-poor performance is secondary. However, to the degree that poor populations have influence on the state of the land condition, it is important to consider the impact on them and to work to align the incentives of all parties involved (Akiefnawati et al. 2010). In cases in which incomes are improved and poverty is reduced, there is an incentive for individuals to assist in providing the service rather than to undermine the project. Ultimately, such assistance reduces the need for extensive external monitoring and helps to reduce transaction costs. The importance of poverty as a secondary goal is greater in situations in which property rights are less clear. It is important that payments compensate users and de jure owners for their opportunity costs; otherwise the owners have incentives to accept payment but do not have incentives to monitor. Coomes et al. (2008) indicate that some programs fail to compensate opportunity costs properly, making them unlikely to succeed. Communities may engage in collective action to subvert the program rather than to support it when they are not sufficiently compensated (Swallow & Meinzen-Dick 2009). If programs do not provide sufficient payment when monitoring is difficult, cheating and leakage will be more prevalent.

2.2.4. Nested institutions

Developed nations typically enjoy lower transaction costs for PES programs because extant institutions are in place and there is greater clarity on land tenure. Due to larger incomes, there is an ambiguous impact on the likely success of a PES program. On the one hand, higher productivity and prices cause opportunity costs to be larger, making the necessary funding amount greater. On the other hand, as the environment is often seen as a luxury good, some conservation may occur naturally. Due to this increased preference for the environment, PES schemes in developed nations may fail to deliver additional ES. Furthermore, even in developed countries, the process of monitoring remains difficult (Claassen et al. 2008, Dobbs & Pretty 2008).

The form of contracts and negotiation can have great influence on the effectiveness of a PES program. One obvious component of an effective program is conditionality. If the service is not provided, payments should cease. But this condition entails both monitoring and enforcement costs. Timing is also important. There are often large upfront costs in conservation, particularly when active land use change rather than passive land use change is required. In poor populations that are financially constrained, even a lucrative long-term conservation deal will fail if early payments do not cover the upfront labor and opportunity costs.

Although PES programs may provide an incentive to develop more secure property rights, the introduction of payments can result in perverse incentives, increasing degradation and crowding out social behavior. Introducing payments into a setting that operates on social capital may undermine prior cooperation. While providing a good review of the experimental literature on this regard, Bowles (2008, p. 1,605) states that the economic policy design often “overlooks the possibility that economic incentives may diminish ethical or other reasons for complying with social norms and contributing to the common good.” In field experiments conducted in Mexico and Tanzania, Kerr et al. (2012) find elements of Bowles's (2008) contention that economic schemes undermine ethical reasons for compliance. Overall these researchers find that low individual payments crowd out participation in communal projects in Mexico and Tanzania, in contrast to situations with no mention of cash. Furthermore, they find that those participating in the communal task for payment are less satisfied with the process than those volunteering their time. At the local level, providing in-kind compensation, rather than incentive payments, may reduce a breakdown in social norms (Van Noordwijk & Leimona 2010, Vatn 2010). García-Amado et al. (2011) find that participants cite income as the third-most-important reason they engage in conservation. Although this aspect introduces questions of additionality and motivation, the other concern is that, in the presence of PES schemes, degradation increases by individuals seeking to establish additionality (Andersson & Richards 2001). Once it becomes clear that marginal land is worth something under a PES scheme only if it will be altered, individuals may begin to modify land use strategically to become eligible. Alternatively, they may resort to blackmail directly, threatening to deforest unless compensated (Karsenty et al. 2013).

Designing a PES that works at various levels can reduce transaction costs by limiting the number of parties negotiating while utilizing local knowledge on how to cost-effectively meet the targets. The nesting of institutions provides this opportunity. Choosing the proper compensation form at the various levels can aid in making a more effective program. Van Noordwijk & Leimona (2010) differentiate compensation into three broad categories: (a) A commoditized environmental services (CES) program provides funds conditional on the service being provided; (b) a compensating opportunities skipped (COS) program derives payments conditional on inputs actually being employed; and (c) a coinvestment in land stewardship (CIS) program operates on trust, with the local users providing often nonmonetary incentives to engage in environmentally friendly practices. According to Hoang et al. (2013), employing CES with cash payments at the top levels, such as the central government to the districts, can be effective. At this level, land ownership is clear, and the district producing the service is clearly identified. Farley et al. (2010) find this technique to work well in the context of Brazil's ICMS ecológico program. Hoang et al. (2013) suggest that, from the district level and below, COS or CIS may be more appropriate to allow negotiation and flexibility at the grassroots level. These programs can overcome the obstacles discussed above such as costly upfront costs. Specifically, by using CIS, one can avoid using money directly and the perverse incentives that it may introduce. Agrawal et al. (2011) point out the importance of engaging multiple levels of actors; the state often maintains de jure property rights but sometimes decides to devolve some management rights to more local levels. White & Martin (2002) estimate that most of the forest in developing countries—79%—is publicly owned, although only 71% is managed by the state. The remainder is private, with 14% held communally and 7% held individually.

2.3. Interaction of Physical and Institutional Attributes

If we draw on Ostrom's (2009) SES framework, attributes of the service will interact with the institutional structure to impact the level of transaction costs and ultimately the likelihood of success. Scholars advocate applying the framework and lessons learned in the context of common pool resources and SES to the PES context (Fisher et al. 2010, Yin & Zhao 2012). These considerations could guide program design to maximize the ES targeted. Ostrom's SES framework is a diagnostic tool for policy makers and needs further hypothesis testing. For example, as mentioned above, the tenure situation is of first-order importance before a PES scheme is established. Another factor is the resource context: PES programs for carbon sequestration may need strategies different from those for programs for other ES. In short, the SES framework is a living framework that needs testing to verify which contexts matter the most. The SES framework can guide research and policy, but it is not a blueprint. Practitioners and academics need to be aware of both the strengths and weaknesses of the SES framework.

One big split in the PES debate is user-funded programs versus government-funded programs. Overall, user-funded programs appear better suited for water services. In these cases, the group of users and providers is easy to identify. In this context, incentives are compatible because those downstream truly desire better-quality water or more reliable sources of water. In this situation, conditionality is likely to be enforced. Because of the limited geographic scope of the service, local groups can utilize their knowledge of the situation to come up with a PES scheme, alleviating the need for a central government to try to impose a system across all the watersheds.¹² Carbon sequestration, in contrast, needs government programs more to handle the public good. Large programs can benefit from economies of scale in transaction costs and provide additional monitoring to ensure quality of the service.

Given the difficulties of directly measuring carbon sequestration and establishing additionality of avoided deforestation, programs can reduce transaction costs by focusing on reforestation. Although such an approach is initially more expensive due to labor and inputs, the question of additionality is alleviated, and land use changes are easy to monitor. Of course, this approach could lead to perverse incentives in existing forestland, making it necessary to assess the overall importance of additionality. In contrast, because location and type of vegetation are important to biodiversity and watershed services, these programs cannot focus only on reforestation, as precise pieces of land, such as wildlife corridors, need to be maintained and improved.

The state of the literature is long on potential hypotheses about the ingredients for successful PES programs. Much more empirical work is necessary to provide generalizable hypotheses for policy makers.

3. PES In Action: Will It Work? Is It Working?

PES programs targeting an assortment of ES have recently been implemented across various settings. We surveyed a selection of studies of PES from around the world, targeting a variety of ES to analyze whether programs succeed in delivering additional ES while considering the extent of the transaction costs. Noticeably, the sample is small, consisting of 18 studies analyzing 16 programs. We also report on an additional 4 studies that use simulation and valuation methods to assess the potential feasibility of PES programs. The small sample is an artifact of the paucity of programs and the difficulty of analyzing them. Researchers struggle with the same measurement issues as program administrators: measuring the flow of the service and establishing additionality. In fact, most studies rely on input measures rather than on output. Researchers try to address additionality in their studies by carefully defining the counterfactual. The sample is not large enough or similar enough to perform rigorous meta-analysis. Instead, we present the techniques used and highlight some key points. Ultimately, most researchers perceive some positive amount of ES, although the extent and efficiency of the programs vary. Table 1 summarizes the empirical studies assessed.

Table 1. Empirical studies^a.

Program	Country	ES	Funding	De jure illegal	Conditionality	Transaction costs	Paper	Additionality	Leakage	Poverty reduction	Covers opportunity cost	Outcome
PSA	Costa Rica	All	Gov.	No	High	15%	Locatelli et al. (2008)	N/A	N/A	No	No	Positive
							Pfaff et al. (2008)	Low	N/A	N/A	N/A	Marginally positive
							Zbinden & Lee (2005)	Low	Possible	No	N/A	Mixed
							Sierra & Russman (2006)	Low	Possible	N/A	N/A	Marginally positive
							Pagiola (2008)	Low	N/A	No	N/A	Positive
Ecotourism	Cambodia	Biodiversity	Gov.	Yes	Yes—output	High	Clements et al. (2010)	Yes	N/A	N/A	High	Positive
Ag certification	Cambodia	Biodiversity	Gov.	Yes	Yes—output	High	Clements et al. (2010)	Yes	N/A	N/A	High	Positive
Monitoring	Cambodia	Biodiversity	Gov.	Yes	Yes—input	Low	Clements et al. (2010)	Yes	N/A	N/A	High	Positive
Biosphere reserve	Mexico	Biodiversity	Gov.	Yes	Yes—outcome	Unknown	Honey-Roses et al. (2011)	Yes	No	N/A	N/A	Positive
Single contract	Tanzania	Biodiversity	User	No	Yes	Low	Nelson et al. (2010)	Likely	No	No	Likely	Positive
CSS	United Kingdom	Biodiversity/landscape	Gov.	No	Yes	18%	Dobbs & Pretty (2008)	Yes	Within farm	No	N/A	Positive
CLP	Mozambique	Carbon	User	No	Low	66%	Hegde & Bull (2011)	N/A	N/A	No	Yes	Positive
							Groom & Palmer (2012)	Likely	Likely	Likely	Yes	Positive
PSA-CABSA	Mexico	Carbon	Gov.	No	Low	30–50%	Corbera et al. (2009)	Yes	N/A	N/A	Yes	Positive
PROFAFOR	Ecuador	Carbon	User	Yes	Yes—lower community	25%	Wunder & Albán (2008)	Yes	Low	Yes	N/A	Positive
PASOLAC	Central America	Water	User	No	Low	High	Kosoy et al. (2007)	Low	Low	Unlikely	No	Positive
PSAH	Mexico	Water	Gov.	No	Yes	4%	Muñoz-Piña et al. (2008)	Low	Some	Yes	N/A	Marginally positive
Pimampiro	Ecuador	Water	User	Yes	Yes—input	17%	Wunder & Albán (2008)	Yes	None	N/A	N/A	Positive
WFW	South Africa	Water	Gov.	Yes	Yes—input	Low	Turpie et al. (2008)	Yes	None	Yes	Yes	Positive
SLCP	China	Water	Gov.	No	Yes—retired	High	Groom & Palmer (2012)	Yes	Potential	Yes	Yes	N/A
							Yin & Zhao (2012)	Yes	N/A	N/A	N/A	Positive
CRP	United States	Water/wildlife	Gov.	No	Yes—low inspection	1%	Claassen et al. (2008)	Yes	20%	N/A	Yes—bidding	Positive

^aAbbreviations: CLP, Carbon Livelihood Project; CRP, Conservation Reserve Program; CSS, Countryside Stewardship Scheme; ES, environmental services; PASOLAC, Program for Sustainable Agriculture on the Hill Sides of Central America; PSA, pagos por servicios ambientales program; PSA-CABSA, Program of Payments for Carbon, Biodiversity, and Agro-Forestry Services; PSAH, Program for Hydrological Environmental Services; SLCP, Sloping Land Conversion Program; WFW, Working for Water program.

3.1. PES Feasibility

Some work has been done to assess the economic feasibility of PES programs. The basic economic criterion necessary is that the willingness to pay (WTP), the monetary utility derived from consuming the service, be greater than the willingness to accept (WTA), the marginal cost of producing the service.¹³ Van Hecken et al. (2012) focus on the demand side, conducting a contingent valuation survey of water users in Nicaragua. Van Hecken et al. find that the WTP is significant; however, users are willing to make greater payments for infrastructure improvements compared with payment to land owners upstream. Börner et al. (2010) find similar results in Brazil concerning carbon sequestration; the economic underpinnings exist (WTP exceeds WTA), but the local institutions are not as conducive to a PES scheme. In contrast, Coomes et al. (2008) find that it is not economically prudent to reforest in Panama given the return on grazing cattle. Looking at the net present value of income streams from both reforestation and cattle grazing over 25 years, Coomes et al. find that reforestation is slightly better but that the timing of the payments makes it difficult for financially strapped farmers to participate. Without access to capital markets, the trade-off between the upfront costs of reforestation and the upfront benefits of cattle grazing is difficult to make, despite the long-term benefits.

Busch et al. (2012) perform simulations of various policies designed to reduce carbon emissions in the context of Indonesia. They consider what would have happened from 2000 to 2005 by calibrating a model of observed deforestation and spatial variation in economic benefits of land conversion. They find that, whereas a mandatory tax/subsidy at a global market rate reduced emissions by 26%, a voluntary PES system cut emissions by only 8%. The voluntary performance improved to 22% reductions when contracts targeted larger plots of land (reducing leakage), monitoring techniques improved, and nested institutions were employed to share both costs and benefits. The cost sharing supports Palmer's (2011) argument that sharing liability between levels will help decrease leakage and increase permanence. Ultimately, the results here bring the effectiveness of PES schemes into question. To assess such effectiveness, we turn to studies that explore individual cases.

3.2. Costa Rica's PSA program

Costa Rica claims one of the earliest implementation of a PES program in the developing world. The pagos por servicios ambientales (PSA) program began in 1997. The program is an umbrella program in the sense that it targets multiple services: The program seeks to improve the provision of carbon sequestration, hydrological services, biodiversity, and scenic beauty, with the government gaining the rights to any carbon sequestration (Pagiola 2008). It is difficult to assess the effectiveness of a particular resource service, but the program's age and longevity have led to a relative abundance of analyses. The various studies provide a good sample of the techniques used to assess the efficacy of PES programs. Overall, the results are not a glowing endorsement of actual performance. Three of the five studies included find negative or, at best, marginally positive outcomes, and the other two find positive outcomes. Notably, additionality is weak across the studies; leakage undermines program efficiency. Targeting multiple services makes it difficult to select the appropriate land to enroll.

The two techniques most commonly used for assessment are econometric analysis of household survey data and econometric analysis of satellite data. Satellite data are seen as a promising technological improvement allowing for lower-cost monitoring, although these data are limited to quantifying inputs rather than output directly. Utilizing econometric analysis, regardless of the data source, seeks to create a more accurate business-as-usual scenario by including similar observations not enrolled in the PES. Locatelli et al. (2008), Sierra & Russman (2006), and Zbinden & Lee (2005) utilize household-level surveys to analyze PSA.

Sierra & Russman (2006) include nonparticipants in the sample and use regressions, controlling for attributes that influence deforestation. These researchers find no discernible difference in forest coverage between participants and nonparticipants. Locatelli et al. (2008) find a positive overall effect, although they limit their sample to those under reforestation contracts, making it easy to assess the change in land use. Using fuzzy multicriteria analysis, they focus on other outcomes, finding that the economic impact is negative and that the poor population does not benefit. Thus, Locatelli et al.'s findings bring the sustainability of the program into question. Notably, though, these scholars find that land tenure became more secure, underscoring the endogeneity of property rights. Zbinden & Lee (2005) use their survey of participants and nonparticipants to highlight the selection bias of participants. The extent to which participants were well off, were better educated, had more off-farm income, and had more secure tenure on larger plots of land largely explains the odds of participating, whereas physical attributes of the land, such as degradation and slope, are less influential. Overall, the evidence brings into question the additionality being achieved.

Pfaff et al. (2008) utilize objective satellite imagery rather than household surveys to analyze the impact on deforestation. To utilize an accurate counterfactual, they use matching techniques based on the physical attributes and access to economic activity. Overall, they find that the land enrolled tended to be at low risk of deforestation on the basis of observables, finding that 99.92% of the land would have been conserved anyway due to lack of market opportunities for much of the enrolled land. Ultimately, this finding implies that only 0.08% of the funds were effective in altering land use changes and that the rest was simply a transfer of funds. Finally, Pagiola (2008) approaches the issue by exploring the literature surrounding PSA, finding that additionality is probably weak but that institutions do improve. For example, property rights to land became more secure. Pagiola also finds that compensating those actually providing the service mattered. Although not every program shares the same lackluster performance in additional service provision, the techniques utilized across these studies provide an excellent idea of the tools used in assessing PES schemes.

3.3. Water PES Programs

Six programs targeting watershed services from the United States, Latin America, China, and South Africa are included in the discussion here. Of the three programs in Latin America, only the Mexican Program for Hydrological Environmental Services (PSAH) is government funded. The authors of the studies attribute this to the regional provision of the public good, making identification of the beneficiaries difficult. As suspected, this program provides the lowest transaction cost of the group: 4% (Muñoz-Piña et al. 2008). In contrast, downstream users fund both the Program for Sustainable Agriculture on the Hill Sides of Central America (PASOLAC) in Nicaragua, Honduras, and Costa Rica and the Pimampiro project in Ecuador. Kosoy et al. (2007) underscore the potential use of PES in watershed services, finding that the users had a WTP that was much higher than the payments they were making. In Ecuador, Wunder & Albán (2008) find the program to be effective and highlight a few distinctions. First, despite the higher transaction costs, the user-funded program appears to be closely monitored; payments are denied due to the lack of provision. Second, Wunder & Albán find low levels of leakage attributed to the nature of the watershed and lack of other land that would offset improvements.

The watershed programs in China and South Africa are organized by the government. The Sloping Land Conversion Program (SLCP) is a large program in China aimed at improving water quality and flood protection by removing sloped land from production and reforesting it. The government involvement comes from the sheer size of the program, which is designed to remove 14.67 million hectares from production. Given the poor population that used this land and no intermediate land use options, the government is relocating populations to off-farm industries. This project has been criticized for its inability to meet its targets in a sustainable manner (Yin & Zhao 2012). Groom & Palmer (2012) note that the poor population has seen an increase in income, begging the question as to why this population did not migrate prior to the program. The authors suggest that this setting is odd due to the authoritarian Chinese government.¹⁴

In contrast, the WFW program in South Africa is seen by many as a success (Turpie et al. 2008). The program is unique in that it bypasses property rights issues by paying for labor rather than paying land owners forgoing production. The transaction costs are low; note that WFW was designed as a poverty relief mechanism to employ the unemployed. Furthermore, the government funded it because the vegetation needing removal should have been removed by the land owners by regulation. However, the statute is not well enforced, and much of the land actually belongs to the state. Somewhat unique is that the provision required the removal of an invasive species, making monitoring of the task much more clear and straightforward.

Finally, the US CRP was designed to reduce soil erosion and subsequent water issues, among other services. Overall, the program works with farmers to convert highly erodible cropland to vegetative cover to reduce soil erosion and thus improve water quality and groundwater recharge. The program also aims to provide some wildlife services. Claassen et al. (2008) find that the program is effective in achieving additionality. The program strives to enroll land that will gain the most impact per dollar by relying on bids that calculate the Environmental Benefits Index (EBI) of each bid, considering the cost of providing the service. Although more expensive upfront, the EBI permits greater cost-effectiveness. Even with the upfront costs of the EBI, the program exhibits very low transaction costs, just 1% of the program's costs, due to the preexisting institutions and the scale of the program. However, there remains a trade-off with these transaction costs: Even in this developed-world setting, monitoring remains quite low. A program this large exerts forces on the crop markets. The decrease in supply forces prices up, inducing others to enter. The study suggests that, for every 100 acres enrolled in the program, 20 acres elsewhere are brought into production.

3.4. Wildlife Habitat and Biodiversity PES Programs

Almost all biodiversity and wildlife programs are government funded. In most cases, these programs target general biodiversity, giving the government a necessary role, as the beneficiary is the public, which, absent government intervention, will underprovide the good due to free riding. The user-funded program is unique in that it is the most Coasian and involves negotiation over a single contract whereby a marketable good correlates with the service. We discuss this project last.

In the developed-world setting (i.e., the United Kingdom), the Countryside Stewardship Scheme (CSS) is a PES program that enrolls landholders to provide wildlife services. Notably, the CSS is the only PES program we found that explicitly targets the provision of landscape beauty (Dobbs & Pretty 2008). The program grew rapidly, and just 13 years after its start, it had entered into more than 16,000 contracts to cover more than 50,000 hectares. Dobbs & Pretty (2008) note that most of the enrolled farmers have been running relatively small scale farming operations, and the CSS has had relatively limited success in attracting highly industrialized agricultural operations to enroll in the CSS. According to the authors, this shortcoming is attributable largely to the availability of competing payments in the form of generous agricultural subsidies from the European Union's Common Agricultural Policy. The authors review the evidence on the extent to which CSS produced environmental additionality and find that approximately 75% of all enrolled farmers carried out conservation activities that would not have taken place, at least not at the same scale, in the absence of the CSS payments. The authors also note, however, the inherent methodological challenges in quantifying such additionality: to determine the value of the resources that would not have been conserved had it not been for the payment.

Clements et al. (2010) explore three PES schemes in Cambodia in a small region. All three schemes are targeted at protecting the habitats of a bird species. Given this specific target, the service is quantifiable, and all three programs resulted in more nests and more birds. Again, the government provides additional incentives here, as all the villages targeted are inside so-called paper parks where the protection should already exist. The findings support the use of a variety of payment schemes, as both sharing ecotourism revenue with the locals and providing a certification for market goods improved the habitats over the long term. The third technique was to pay individuals to guard the nests from poachers; this technique was effective immediately, although long-term sustainability is questioned. This approach does reaffirm the cheap, low-transactioncost option to include those without clear property rights and that people respond to economic incentives more than to regulatory constraints. The use of payments for labor can complement payment to holders of property rights, including a greater number of locals, while providing greater odds of success (García-Amado et al. 2011). Not only does the extra payment reduce the non–property rights holders' need to illegally harvest forest products, but such individuals also have an incentive to help enforce the policy. Aligning the incentives of more users, both de jure and de facto, increases the odds of success. The inclusion of labor payments has effectively done so.

Honey-Roses et al. (2011) employ satellite imaging and illustrate the need to construct an accurate counterfactual. In assessing the impact of paying communities to protect the Monarch butterfly in Mexico, they match plots of land on the basis of physical characteristics and consider what is occurring on adjacent plots. Without this spatial consideration, they find no improvement in forest coverage, but once they match observations to create a baseline of potential deforestation, they find evidence of additionality. Although this finding highlights the need to consider the business-as-usual scenario for avoided deforestation, it does not indicate that doing so will create a more positive outcome. These results contrast with those of Pfaff et al. (2008), who find no additionality. Controlling for potential deforestation, Pfaff et al. (2008) find the impact of the PSA program to be small, indicating that land under little stress was more likely to be enrolled in the program. We highlight that additionality results are likely dependent on the setting, but both examples illustrate a need to consider the appropriate counterfactual to establish additionality.

The outlier of the studies looks at a single contract in Tanzania between a user and a provider (Nelson et al. 2010). The contract represents a quintessential example of a Coasian-type negotiation in which clear property rights aid in contracting. Unlike in the other cases of biodiversity, this service provided a direct financial benefit to tour companies, creating a quasi-private good. Unlike other user-funded programs, the transaction costs were low. This should be of no surprise, for if they had not been, the deal would not have been sought. Preexisting relationships appear to have kept down the negotiation costs. Tour guides in the area became concerned that land owned by the village, which was historically undeveloped, was beginning to be cultivated and settled. The region serves as an important grazing land for the animals that the tours like to showcase but is not directly accessible to tours. This characteristic distinguishes the contract from other concessionary agreements that tour companies have with villagers to take tours on their land. However, given these prior agreements, the new negotiation built on prior interactions of the contracting parties. The tour guides agreed to pay the villagers a set amount each year to not cultivate or to establish a permanent settlement. The deal included additional funds to pay for monitoring and guards. Although no count on the increase in animals is available to truly assess the environmental impact of the contract, the land has maintained its natural state, and the funds have been used to prosecute invaders from other villages; both outcomes indicate a change in behavior.

3.5. Carbon PES Programs

The three programs included in the study that assess PES programs for carbon sequestration have notably high transaction costs. The Carbon Livelihoods Project in Nhambita, Mozambique, stands out, with transaction costs estimated to constitute 66% of the total costs (Hegde & Bull 2011). The evidence confirms our ex ante expectations of carbon sequestration programs having the largest transaction costs due to the pure-public-good nature and to difficulty in monitoring and enforcement. A paper by Hegde & Bull (2011) reports that the program decreased agricultural activity and deforestation on the basis of surveys of participants and nonparticipants, noting that households see an increase in income and their opportunity costs covered. Furthermore, a paper by Groom & Palmer (2012) offers some evidence of additionality despite the transaction costs. Both papers attribute some of the success to the menu of options available to landholders, allowing them to pick the technique of provision most suitable to their needs. There is concern that, because of the large transaction costs, the providers are not receiving enough of the funds to make the program sustainable.

The other user-funded carbon program considered is the PROFAFOR project in Ecuador (Wunder & Albán 2008). This program is also seen as a success in terms of additionality and lack of leakage. The Dutch company seeking carbon credits monitors the individual plots of land each year to measure forest coverage.¹⁵ As this program is user funded, there is some incentive to ensure that the service is provided. To reduce leakage and transaction costs, the program allows only plots of 50 hectares or more to be enrolled. The Dutch also consider biophysical attributes to target plots that would provide more additional sequestration.

The last program—Program of Payments for Carbon, Biodiversity, and Agro-Forestry Services (PSA-CABSA) in Mexico—is government funded. Thus far, it appears to have had some positive impacts, although the extent and continuation are not clear (Corbera et al. 2009). Funding has dropped off considerably. The program, although supposedly conditional, has exhibited lackluster monitoring. In contrast to the user-funded programs, it appears that government-financed programs are less likely to ensure the provision of the service and may be implemented for political gain rather than for environmental outcomes.

3.6. Broad Themes

Transaction costs that are classified as low tend to be associated with central government–pfunded programs. As discussed above, the transaction costs of the US CRP program are only 1% of the program's overall cost. This figure is extremely low compared with the 66% figure in the Nhambita project. As one of the few programs explored in a developed-world setting, the CRP program has a relatively low transaction cost that is not surprising due to the extant institutions that often accompany development. For example, extensive data on farms and land use in the United States predate the CRP, providing a baseline with which to assess performance. In addition, the scale of the project keeps down the transaction costs as a percentage. In 2004, the program expended $1.85 billion, of which $15.5 million was for transaction costs (Claassen et al. 2008). However, the generalization of transaction costs as high for users and low for governments does not hold completely across our sample. For instance, the SLCP in China is government run and has high transaction costs, although this is a very large program lacking the infrastructure existing in the United States. In contrast, there appears to be a positive relationship between transaction costs and the public nature of the good. That is, transaction costs are lowest when the programs target club goods, such as watershed services, and highest when targeting a global public good such as carbon sequestration.

Another important issue is the role of property rights. Notably, many projects exist where the service is already legally required. The necessity to pay users to adhere to the law highlights the potential use of payments rather than sanctions because the current institutional structure does not provide proper enforcement of the existing laws.¹⁶ The places where users are financing the program exhibit the sufficiency of de facto rights, as users are paying for a service that is required by law (Wunder & Albán 2008). Some programs, e.g., PSA-CABSA and PSAH in Mexico (Corbera et al. 2009, García-Amado et al. 2011, Muñoz-Piña et al. 2008), allow communal land to be enrolled. Other programs, e.g., the US CRP and Costa Rica's PSA, also allow those with incomplete or insecure property rights to participate (Claassen et al. 2008, Zbinden & Lee 2005). The extant evidence indicates that full private property rights are not necessary for successful PES programs and that communal and incomplete rights are sufficient in many cases.

4. REDD+ Implications

REDD+ offers an alternative method to reduce carbon levels by increasing the capacity of the socalled sink to absorb carbon rather than by reducing actual emissions. As discussed above, the issue with carbon sequestration is that it is a pure public good, leading to the underprovision of the service. However, it is estimated that 1/5 of global emissions of greenhouse gas are attributable to forest degradation, making such degradation important to consider in policy discussions (Palmer 2011). To provide incentives for nations, and ultimately communities and individuals, to reduce their carbon footprint, certain institutional structures must be present to create effective provision of the good. The length of contracts is also important because, although a unit of carbon not emitted is forever out of the atmosphere, one that is sequestered will eventually return to the atmosphere.

4.1. Exclusion Creation

Capping emissions and creating permits create an instrument tied to carbon emissions that are now rival and excludable (Farley et al. 2010). Alternatively, developed countries could enact a carbon tax. So long as offsets are permitted, the result will be the same, and REDD+ will allow developing countries to provide emission reductions through land management. Economically, increasing the source of possible emission reductions will not make the program more expensive and may make it more cost-effective.

4.2. Incentive Compatibility

Along with concerns of additionality, conditionality and monitoring are particularly large issues in the international setting of carbon. This is due largely to the nature of the excludable and rival certification. Only the piece of paper that gives credit for emissions reduction is valuable to the purchaser (Alston & Andersson 2011, Swallow & Meinzen-Dick 2009). Unlike in user-funded programs in watershed services, in which the service is actually desired, the carbon reduction itself remains a public good.¹⁷ In this sense, the purchaser of the service has no incentive to report the provider's failure to deliver the service. Naturally, the provider has no incentive to report the noncompliance and forgo the monetary compensation. Thus, the contracting parties have little interest in monitoring the activity and ensuring the service is provided. Even if the carbon rights are decoupled from the land and transferred to the buyer, along with the liability of insufficient provision, thus introducing some incentive to monitor activities, without third-party verification the incentive structure remains incompatible with delivery of additional carbon sequestration.

4.3. PES Mechanisms

PES is a popular mechanism by which to achieve the goals of REDD. Given that offsets are allowed in the regulated nations, positive incentives are the only way to induce nonregulated countries to participate. One nation cannot regulate another due to sovereignty issues. Furthermore, although it is feasible for a nation to purchase land in another country and to have the right to reduce emissions itself, such a step is often viewed as a threat to sovereignty (Vatn 2010). However, payments can provide positive incentives that lead landowners to engage in conservation now that they receive the benefits for the service they are providing. For this reason, policy makers view PES favorably for implementing REDD+. However, there remains considerable debate as to whether these PES schemes will provide actual carbon sequestration and whether the transaction costs involved will erode the mutual benefits.

Briefly returning to empirical results, we find that the evidence is not favorable toward effectiveness. PES schemes have not achieved much additionality of carbon sequestration, and in some cases there is little evidence that land use has even changed. The large hurdles are unclear property rights and large transaction costs stemming from costly negotiation (not getting de facto and de jure rights holders at the table) and monitoring and enforcement costs. Users funded a majority of the carbon PES schemes included in our study, demonstrating that policies creating exclusionary goods are somewhat effective. However, such schemes are also plagued by high transaction costs, and it remains unclear that sufficient monitoring and verification will occur.

4.4. REDD+ Policy

To achieve true reductions in a cost-effective manner, REDD+ PES schemes can be informed by theory and empirics. Nesting levels of provision is one avenue that can increase the additionality while providing more effective monitoring and competition to reduce costs. By negotiating at the nation level, governments can economize on the transaction costs. At this level, conditionality can be on actual aggregate performance. This process still needs to be verified by an independent third party. However, once this criterion is established, monitoring at the lower levels is incentive compatible. The state has the incentive to allocate the funds to successful projects. This competition for funds by local and regional governments can also provide cost-effective mitigation, similar to the benefits of an auction in alleviating information asymmetries. The nested approach also eliminates tenure issues at the international level, as a nation certainly has rights to its own land. The nation state may in turn assign property rights and monitoring to lower subnational and local units of governance, which should reduce overall transaction costs. Potentially, REDD+ could provide further benefits by giving governments of developing countries incentives to establish the institutions necessary to deliver the service (Agrawal et al. 2011). Institution building in turn could be a public good for the developing countries. Even if nestedness is not utilized, targeting communal lands or large private owners is important. Larger portions of land provide savings in transaction costs and reduce the possibility of leakage.

5. Conclusion

PES is subject to many transaction costs. We expect these to increase when one is moving from watershed services to biodiversity protection to carbon sequestration mainly on the basis of physical attributes of these services. The existing empirical literature confirms our prediction. Although our analysis does not look favorably on the effectiveness of a PES-type scheme to initiate REDD+, we present some steps that could reduce transaction costs and increase the probability of success. More work needs to be done to assess whether parties engaged in REDD+ initiatives foresee the downstream transaction costs and negotiate ex ante to lessen the otherwise opportunistic behavior.

We also stress that the bulk of PES programs are relatively new and that with experimentation and learning the results should improve. For now, the jury is still out on the effectiveness of PES, yet there is reason for optimism over time, especially with regard to local club goods. In addition, PES programs should not be viewed in isolation from other programs for sustainable management. Indeed, we encourage the environmental community to consider PES programs as complements rather than as substitutes for other programs aimed at sustainable management.