PES Learning Paper 2015-1 Do They Do As They Say? Stated versus Revealed Preferences and Take Up in an Incentives for Conservation Program Samantha De Martino, Florence Kondylis, Stefano Pagiola, and Astrid Zwager December 2015 (rev. June 2016) Environment and Natural Resources World Bank Washington DC, USA Abstract Use of conditional cash transfers has become widespread in development policy given their success in boosting health and education outcomes. Recently, conditional cash transfers are being used to promote pro- environmental behavior. While many of these Payments for Environment Services (PES) programs have been successful, it has been hypothesized that those with less favorable outcomes have been subject to low additionality, whereby landholders already conserving their land self-select into the program. Insights from the behavioral economics literature suggest that an external incentive, such as PES, has the potential to crowd in or crowd out individual behavior differentially across the initial distribution of intrinsic motivations (Frey, 1992). Thus, to increase the impact of PES, program administrators might gain from a better understanding of both the pre-existing motivations and existing baseline conservation behavior of potential participants. This paper contributes to the literature by disentangling and measuring intrinsic motivations, specifically: Pro-Environment, Pro-Social, Pro-Government, and Social Norms. Controlling for observable opportunity costs, we use these latent motivations to analyze behavioral determinants of take up for a conservation program in São Paulo, Brazil. The payments are an incentive to comply with the Brazil Forest Code. We find that Pro-Social and Pro-Environment landholders are both more likely to be conserving private land not under legal protection before the program is introduced, whereas only Pro-Social landholders are already conserving land under legal protection. With respect to enrollment in the PES program, we find Pro-Social landholders are less likely to enroll while Pro-Environment landholders are more likely to enroll. Thus we expect some level of additionality from the PES program. We discuss these findings in light of the theoretical framework on Self-Determination Theory (SDT). Authors Samantha De Martino is a PhD student in the Department of Economics at the University of Sussex; Florence Kondylis is Senior Economist in the Development Impact Evaluation Unit, World Bank; Stefano Pagiola is Senior Environmental Economist in the Environment and Natural Resources Global Practice, World Bank; Astrid Zwager is Consultant in the Development Impact Evaluation Unit, World Bank. Keywords Payments for Environmental Services (PES); conservation; motivation to conserve; crowding out; intrinsic motivations, extrinsic incentives; self-determination theory. Acknowledgements This paper is based in part on work carried out during implementation of the São Paulo Sustainable Rural Development Project, implemented by the São Paulo State Secretariat of Environment’s (SMA) and State Secretariat for Agriculture (SAA) and funded by a US$78 million loan from the World Bank. Support was also received from the World Bank Research Committee and the Spanish Fund for Latin America. We would like to thank the SMA’s Coordination Unit for Biodiversity and Natural Resources (CBRN), and in particular Helena Carrascosa, Araci Kamiyama, Caroline Vigo Cogueto, and Ana Carolina Dalla Vecchia; Marianne Grosclaude of the Agriculture Global Practice, World Bank; and Professor Dale Whittington of the University of North Carolina. We are also very grateful to two anonymous reviewers for their constructive feedback. Cover photo Farm in Guapiara municipality (Stefano Pagiola). PES Learning Papers PES Learning Papers draw on the World Bank’s extensive experience in supporting programs of Payments for Environmental Services (PES). They are part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. The PES Learning Paper series disseminates the findings of work in progress to encourage the exchange of ideas about PES. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Do they do as they say? Stated versus revealed preferences and take up in an incentives for conservation program Samantha De Martino, Florence Kondylis, Stefano Pagiola, and Astrid Zwager 1. Introduction Conditional cash transfer (CCT) programs have become common in development policy due to their success in boosting health and education outcomes (see, among others: Fiszbein and others, 2009; Gertler, 2004; Skoufias and others, 2001). Payments for Environmental Services (PES) are similar in concept, offering cash payments for conservation (Ferraro and Kiss, 2002; Wunder, 2005; Pagiola and Platais, 2007; Engel and others, 2008; Wunder and others, 2008). PES is subject to the same question as CCTs, namely: do such programs actually motivate people to change their behavior? Due to the self-selection nature of many incentive programs, there is a risk that the incentives go to those who already comply with the program’s conditions, thus limiting the program’s additional conservation impact on the environment (“additionality”). Despite an increased use of PES in the field, the analysis of such programs have received less attention in the economics literature than other types of CCTs (Pattanayak and others, 2010). The impact of these programs has significant fiscal and environmental implications, which are relevant for policy making. Of the few impact evaluations of PES programs that have been conducted to date, the results are mixed. While some programs have had high impacts (Alix-Garcia and others, 2015; Pagiola and Rios, 2013; Pagiola and others, 2016; Arriagada and others, 2012), others appear to have had limited impacts (Ferraro and Pattanayak, 2006; Pattanayak and others, 2010; Robalino and Pfaff, 2013). In cases where the impact has been limited, the hypothesized mechanisms are (i) low or no additional conservation (“additionality”) because participating landholders would have conserved their land even in the absence of the incentive program (Sierra and Russman, 2006; Sills and others, 2008; Pagiola, 2011; Robalino and Pfaff, 2013); and, (ii), “slippage” or “spillovers”, whereby deforestation is diverted to areas not covered by the program (Alix-Garcia and others, 2012).1 An additional explanation for low conservation impact—which is yet untested— is that offering payments for conservation might undermine intrinsic motivations to conserve. This question can be divided into two: (i) whether landholders in fact have intrinsic motivations to conserve, the basis for these motivations, and if these motivations drive participation in PES programs; and (ii) whether participation in PES programs undermines or supports these motivations. In this paper, we use data from a PES program being implemented in the Brazilian state of São Paulo to examine the 1 There are also concerns that impacts may not be permanent (Pagiola and others, 2016). Lack of permanence would not affect results at the end of the program, however. 1 first of these questions (examining the second will require waiting for the project to be completed in order to collect end line data). In principle, PES seeks to change the land use behavior of land degrading property owners for the benefit of the community. The assumption behind PES is that intrinsic motivations are less important than economic incentives in determining behavior. However, the effect of a payment on behavior may vary across individuals depending on their pre-existing intrinsic motivations and what triggers those motivations. Insights from Self-Determination Theory (SDT) suggest that the outcome is dependent on how payments satisfy not only the landholder’s need for profit but also his need for purpose and self-satisfaction (Deci and Ryan, 1991; Ryan and Deci, 2000). Specifically, the decrease or increase in intrinsic motivation is due to three psychological factors: how the incentive affects the need to feel competent, the need to be self-determined, and the need to feel connected to others. Ezzine-de-Blas and others (2015) extend SDT to also include the need to feel connected to the environment. Depending on how the incentive triggers these psychological factors, the incentive may feel imposed on the individual and result in a crowding out of their motivation; or alternatively, the incentive may be internalized if it triggers self- satisfaction and crowd in their motivation. The outcome may determine whether the landholder enrolls in the program. The assumption behind the crowding out literature is that individuals (i) have pre-existing intrinsic motivations, and (ii) that these motivations can be triggered by an external incentive. If we assume that PES contract designers have asymmetric information, then gathering information on pre-existing motivations of potential PES recipients might better inform the principal prior to both parties signing the contract. Landholders who are already intrinsically motivated to conserve according to program conditions may not be the preferred participants in a PES program if the program’s goals are cost-effectiveness and high additionality. Furthermore, to determine if landholders who are not yet conserving their land are incentivized by the introduction of payments, an analysis of how motivations interact with the incentive might help explain take up patterns, which can be used to improve program targeting so as to achieve additional conservation. The intention of capturing intrinsic motivations is grounded in the hypothesis that PES program take-up is not fully explained by observable proxies for opportunity costs. Thus by capturing and measuring motivations, we can obtain more information on the landholders. We can use this information to test the mechanisms through which intrinsic motivations interact with an incentive to participate in a conservation program. However, intrinsic motivations are latent and thus difficult to measure. This paper contributes to the literature on incentives and motivation in the context of PES program take-up by providing tools to understand how to disentangle and measure latent motivations; by testing the validity of the measured latent motivations on revealed preferences of conservation; and by using the latent motivations to analyze determinates of take up for a PES program. 2 Behavioral analysis using survey responses relies on stated preferences, which may or may not correspond to revealed preferences (for example, see Nolan and others, 2008). In the baseline survey for the PES program, we capture stated preferences through an exhaustive survey asking questions on the role of society and individuals with respect to the environment. We conduct factor analysis on these stated preferences to elicit latent motivations, which include Pro-Environment, Pro- Social, Pro-Government and Social Norms. We use these latent motivations as the “stated preferences” in this paper. To overcome the weaknesses inherent in the analysis of stated preferences, we also examine revealed preferences by documenting the pre-existing level of conservation on each property. We then use this data to first examine the role of motivations as a determinant of conservation behavior by studying if the indices constructed from the stated preferences predict revealed preferences, as captured by observed conservation behavior. We hypothesize that landholders with Pro-Environment motivations will already be conserving land that is not under legal protection,2 controlling for observable opportunity costs. Confirming these assumptions, we find that Pro-Environment and Pro-Social landholders are significantly more likely to conserve land not under legal protection before the program is introduced when controlling for a comprehensive set of demographic, socioeconomic and land characteristics. With validation that the stated preferences in our sample capture intrinsic motivations, we explore the extent to which the interaction of these motivations with a monetary incentive affects participation in the PES program. We find that Pro- Environment landholders are more likely to participate in the program. Social norm and Pro-Social motivated landholders, on the other hand, are less likely to participate. Bridging the stated preference analysis with the take up analysis, we find that Pro-Social landholders and Pro-Environment landholders are more likely to be conserving private land outside of legal protection, but only Pro-Social landholders are more likely to be already conserving land under legal protection before the introduction of the payments program. The finding that Pro-Environment landholders are more likely to enroll in the program demonstrates that offering an incentive to those close to the margin of conserving land within legal protection helps to crowd in intrinsically motivated landholders. These landholders are the preferred recipients as opposed to Pro-Social landholders whose enrollment would not result in additional conservation. However, to achieve high additionality, program administrators would benefit most if those with high opportunity costs were motivated by the incentive; we find these landholders are in fact less likely to enroll in the program. We discuss all findings in light of the SDT theoretical framework. 2 “Legal protection” refers to land demarcated as “Areas of Permanent Protection” (APPs), which requires 100 percent conservation. In the case of springs, the APP consists of a 50 m radius from the center of the spring. 3 The rest of the paper is structured as follows: Section 2 describes the context and Section 3 describes the data collection; Section 4 discusses the construction of stated preference indices; Section 5 describes the estimation strategy; Section 6 discusses the results; and Section 7 concludes. 2. Context The Mina d’Agua (MdA) pilot program is being implemented in 21 municipalities, one in each of São Paulo’s Hydrographic Water Management Units (UGRHI) (Carrascosa and others, 2012). All participating municipalities have a legal framework that allows payments to landholders for environmental services. This is the first PES pilot implemented directly by the São Paulo state government. It falls under the state policy on climate change adaptation and mitigation.3 Figure 1: Location of the Mina d’Àgua pilots in São Paulo, Brazil The objective of the MdA program is to preserve and improve the water quality by incentivizing upstream landholders to maintain and/or recover trees surrounding springs on their private property in critical watershed areas. These areas are already protected by the Brazil Forest Code, and are referred to as Areas of Permanent Protection (henceforth “APPs” or “within legal protection”). Hence the payments are 3 An earlier pilot was implemented in collaboration with The Nature Conservancy (Padovezi and others, 2012). 4 an incentive to comply with the existing law. However, in addition to conserving 100 percent of land demarcated by the government, the full Brazil Forest Code also obliges the landholder to conserve 20 percent of land outside of demarcated areas. The MdA program does not impose the second requirement of the full Brazil Forest Code. Priority watershed areas for conservation and restoration are defined as those watersheds that drain to the intakes of local water companies. The project will pay landholders for sustainable land use practices that protect or regenerate forest coverage in a 50m radius around these critical water springs. As mentioned, the PES program does not require any additional conservation and/or restoration. Payments are for up to four springs per landholder and can range from R$74-R$300 per year per spring. The payments are based on a formula that weights the volume of water from the spring, its location, and the degree of conservation. It requires a change in behavior, unless landholders are already conserving their springs, and potentially a loss of income depending on their current land use practices. 3. Data collection A baseline survey was carried out in two of the 21 participating municipalities, Guapiara and Ibiúna. It covered all potential participants (landholders with springs in the priority areas) as well as landholders in areas adjacent to the priority areas who were surveyed as controls for a future end line evaluation of the Mina d’Água program. Data was collected on the landholder’s demographics, socio economic characteristics, characteristics of each individual plot on their property, cultivation and pasture practices, costs and revenues of land use, characteristics of the water springs on their property, willingness to accept compensation for conservation or restoration, and perceptions on the environment. Figures A1-A3 in the Appendix illustrate conditions at the Guapiara study site. We carry out the revealed preferences analysis of conservation (Models 1 and 2, described in the next section) using the total pooled survey sample of 350 households. This comprised of households eligible for the program (211 households) and ineligible households (140 households). We then use the sub level data on eligible households (211 households) to study take up of the program4. In total 107 out of the 211 eligible households enrolled in the program. The average age of landholders responsible for agricultural activities is 55 years, and most landholders live on the property. On average, the landholders have two springs on their property, and over 75 percent of the forest on the property is 4 During the rollout of the project, all farmers within the priority areas with at least one spring were visited by the municipal implementing agency. Extensive information was provided regarding eligibility criteria, conditions and general compensation during these visits. Conversion from interested farmers into project participants likely suffers from some degree of non-random attrition due to misunderstanding of the conditions and benefits. Similar to analysis of potential take-up between control and treatment areas we analyze the intent-to-treat households (those eligible who expressed interest). 5 within APP boundaries. The landholders generally use the springs for both family and livestock consumption. Most landholders do not use credit; the majority indicated that they did not need it, and did not select potential reasons for not using credit (such as lack of access, lack of a guarantor, fear of debt, bureaucracy, or failed payments on other loans). Thus we are likely analyzing a sample population that is either not credit constrained or not undertaking investments. Balance tests compared landholders who enrolled in the program to those who chose not to enroll (see Table A1 in the Appendix). The two groups are generally similar: the average property size is 12 ha, with an average of two agricultural workers. The proportion of steep land in properties in each group is high (77 percent). Approximately 80 percent use land primarily for agriculture. Landholders who did not enroll are currently conserving 71 percent of APP land and 33 percent of non-APP land, while those who enrolled conserve 77 percent of APP and 38 percent of non-APP land. T-tests of means were used to compare landholders from Guapiara and Ibiúna. When comparing descriptive statistics disaggregated at the municipal level, we see significant differences in multiple variables. Therefore, we cluster the standard errors at the municipal level and use municipality fixed effects in all regressions (see Appendix). 4. Capturing Motivations The strength of the baseline survey lies in the exhaustive list of questions on perception of the environment and society’s role in protecting the environment. While we acknowledge that there is a large strand of existing literature on survey methodology to capture pro-environment attitudes (see Clark and others, 2003), we adopted our questions from a previous survey developed by the São Paulo State Secretariat of Environment (SMA) to ensure compatibility with SMA’s surveys in other municipalities. The customized set of questions for participants were context-driven and thus relate directly to the water quantity and quality supply issues in the region5. Pro-Environment motivation Pro-Environment landholders are those who are intrinsically motivated to protect the environment. As a partial proxy to capture this motivation, we explored responses to questions on whether they believed land-degrading activities of their neighbor cause environmental harm (see Appendix for a set of preference questions). We further queried landholders on their perception on the long-term supply of water. We asked if landholders believed there were any benefits to restoration and, if so, what they were (see Figure 2). Improvements in the quality and supply of water were most commonly identified as the largest benefit of restoration. However, strong heterogeneity of beliefs exists in the sample, as the second most popular response was “no benefits”. 5 The full survey is available from the authors. 6 Positive effects Reduces erosion Improves air quality Guarantees water supplies Improves water quality Reduces the costs of conservation Preserves the environment Generates sustainable income None Other Negative effects Reduces income Reduces productive area Takes up too much of small farm None Other 0 50 100 150 200 250 Number of households Figure 2: Perceived benefits and costs of restoration Pro-Social Motivation We asked respondents who is responsible for protecting the environment for future generations. “Everybody” was the most frequent response, which conveys a strong sense of community-driven protection and social connection as defined in SDT (Figure 3). Multiple questions were also asked regarding who is responsible to protect and pay to protect water resources on private and public land (see Appendix). Government Community Landholders Everybody Other 0 10 20 30 40 50 60 70 80 90 100 % of households Figure 3: Household perceptions of responsibility to protect the environment for future generations Social norm indicators Social norms differ from Pro-Social motivations in that the landholder gives weight to the enrollment decision of their neighbors when making their participation decision. This preference may also be viewed as a collective action motivation. In the survey, landholders were asked whether they would change their decision to enroll in 7 the program if they discovered their neighbors had enrolled. Most responded “No.” Another question asked whether landholders planned to discuss this particular project with their neighbors after the survey before making their decision. The majority responded that they would do so, highlighting the importance of peer effects in the community (see Figure 4). Change enrolment behavior if neighbor participates Discuss project with neighbor 0 50 100 150 200 250 300 350 Number of households Figure 4: Effects of neighbors Mechanisms Driving Take Up In the conceptual framework detailed in the introductory chapter of this Special Issue, Ezzine-de-Blas and others (2015) use SDT to explain the channels through which changes in motivations are caused by the introduction of PES. We attribute the four psychological factors outlined in their paper to the intrinsic motivations of interest in this paper, as shown in Table 1. Constructing indices As there are less than 350 observations, we reduce the dimensionality of the explanatory variables by creating indices that capture the various motivations of interest. We consider landholders as having intrinsic motivations: Pro-Social, Pro- Environment, Pro-Government, and Social Norms.6 The focus on these motivations is driven by the desire to disentangle social preferences and understand the main intrinsic drivers of PES uptake. We create composite indices to capture latent preferences. We hypothesize that sets of variables in the baseline survey capture the different motivations. Each index was created by first taking all variables from the baseline survey that could possibly measure the preference and perform factor analysis on them, otherwise known as latent variable analysis.7 Specifically we hypothesize that separate groups of 6 These categories are not necessarily mutually exclusive. However, we tested equality constraints to determine if the motivations have equal effects on the outcomes of interest. We reject that the parameters are equal. 7 This method analyzes observed variation and covariation among observed variables. We used the unbiased Barlett method rather than the default regression method for the purposes of a smaller MSE. 8 Table 1: Effect of PES incentive on landholder’s need satisfaction Psychological Intrinsic moderators Hypothesized channel of how the landholder’s needs satisfaction Motivation triggered are modified by a PES incentive Pro-Social Social If the community values conservation irrespective of payment, the connection payments modify the needs satisfaction by increasing the quality of the landholders’ relations with others in the community. The incentive may have a reverse effect, however, if accepting a payment for conservation is against social norms in the community. Self- The incentive may support self-determination if landholders are determination acting in accordance with their free will to conserve the environment. The incentive may decrease self-determination if compensation is viewed as a control mechanism. Pro- Environmental The incentive may reward the desire to interact and connect with Environment connection the environment. Self- The incentive may support self-determination if landholders are determination acting in accordance with their free will to conserve the environment. The incentive may decrease self-determination if compensation is viewed as a control mechanism. Competence The incentive may enhance competence if the incentive is interpreted as positive feedback on landholders’ existing conservation behavior. Alternatively, it may undermine competence if landholders interpret the incentive as a tool the government finds necessary to increase its existing conservation efforts. Pro- Self- The incentive may enhance self-determination if land conservation Government determination is in accordance with their free will and landholders willingly accept compensation from the government. Competence The incentive may enhance competence if landholders believe the government is rewarding them for their ability to achieve positive environmental outcomes (conservation). Social Norms Social The landholders’ participation decision is dependent on their connection valuation of how conservation behavior, and how accepting a monetary reward in return for positive conservation outcomes, is judged by their community. Informationa Self- The incentive may enhance self-determination through the channel determination of self-development by learning the legal requirements and context. Their participation decision is based on their own intrinsic valuation of costs and benefits. Notes: a. Although not an “intrinsic motivation”, we consider access to full information outside of the strict rational choice model of decision making. 9 variables capture: (i) preferences for protecting the environment (protecting the environment from current land degrading activities, the costs and benefits of restoration, and protecting the environment for future generations); (ii) preferences for who is responsible to protect and pay to protect water sources on public and private property; (iii) attitudes towards the government’s role in environmental protection; and (iv) social norms in the community. In order to test our hypotheses, we use confirmatory factor analysis as opposed to exploratory factor analysis, the former of which is used when the researchers have a pre-defined idea of the structure on a set of variables they want to test. After taking the groups of variables we assume to capture the latent motivations, we first perform factor analysis and then rotate the factors using the default varimax rotation to produce orthogonal factors. Where correlations were high, we use oblique rotations of the factor dimensions. We follow the Kaiser Rule, retaining factors with an eigenvalue cut-off of 1. If there are multiple factors with an eigenvalue higher than 1, then the set of variables are measuring not just one latent motivation, but multiple dimensions of the latent motivation. The main factor analysis is presented in Table 2, and all other factor analyses are included in the Appendix. Taking the factors with an eigenvalue greater than 1, we use the variables which had high factor loadings in the expected direction (see shaded cells in Table 2), and then aggregate by summing the variables and standardizing the composite variable to mean 0, standard deviation of 1 to create the indices. In total, three factor analyses were conducted after systematically reviewing the survey to include all questions that could potentially reflect the motivations of interest. The main factor analysis used to create the latent motivations used in the quantitative analysis—shown in Table 2—has multiple factors with eigenvalues higher than 1. Pairwise correlations were all under 20% for the indices created by the main factor analysis and thus we chose to include each of them in the regressions as they represent unique latent motivations. However, we observe high correlations when we compare them against the indices created by the remaining two factor analyses (see A2 and A3 in the Appendix). For example, “Pro-Social” is highly correlated (=0.97) with the index created from the second factor analysis (A2 in the Appendix). F-tests and likelihood ratio tests conclude us to restrict our model to include only “Pro- Social”. When we include the index from the third factor analysis, f-tests and likelihood ratio tests force us to restrict our model to include only the “Pro- Government” index. Thus we only use the main factor analysis. The questions used to create the indices are included in Table 2. 10 Table 2: Factor Analysis I (rotated factors) Factor 1 Factor 2 Factor 3 Factor 4 Responses to survey questions Pro Env 1 Pro-Social Pro Env 2 Pro-Govt Recovering areas around rivers and springs with forest using public resources would: Improve water quality 0.043 0.023 -0.113 -0.293 Conserve the environment 0.080 -0.023 0.036 -0.028 Improve air quality 0.011 -0.019 0.087 0.991 Improve air quality and conserve environment 0.068 -0.031 0.095 0.760 Improve the environment 0.077 0.002 0.097 0.200 Forest surrounding springs is very important for water quality and quantity 0.005 -0.216 0.078 0.121 Deforestation by your neighbor causes harm 0.053 0.011 0.485 0.042 Damming by your neighbor causes harm 0.081 -0.056 0.825 0.112 Trash disposal by your neighbor causes harm 0.006 -0.137 0.269 -0.052 Use of toxins by your neighbor causes harm 0.064 -0.003 0.327 -0.002 Burning land by your neighbor causes harm -0.006 -0.075 0.261 0.040 All of the above cause harm 0.051 -0.016 0.860 0.106 Water supplies (whether considered abundant, sufficient, or insufficient): Will not diminish -0.997 0.035 -0.044 -0.021 Will diminish 0.997 -0.035 0.044 0.021 Recovering areas around rivers and springs with forest using public resources 0.015 -0.048 0.094 0.124 would have no negative effects for landowners Understands Brazil’s Forest Code and APPs 0.063 -0.047 0.071 0.077 Landholder is highly educated -0.043 -0.074 0.022 0.036 Responsible for protecting the environment for future generations lies with: Community, landholders, or everybody 0.061 0.085 -0.024 -0.018 Landholders only -0.048 0.989 -0.024 -0.014 The community only 0.034 0.105 -0.005 -0.021 Community and landholders -0.028 0.947 -0.024 -0.022 Responsible for protecting water sources in public areas: Community 0.008 -0.048 0.093 -0.081 Eigenvalue 2.035 1.985 1.98 1.762 Proportion of Variance 0.161 0.157 0.156 0.139 Notes: Pro Env 1: Concerned about future supply of water; Pro-Social: Concerned for future generations; Pro Env 2: Concerned about negative impacts of land degrading activities; Pro-Govt: Favors public financed restoration Shaded cells show within each factor, those with high factor loadings in the same direction, and thus included in an index together. 11 Table 3: Correlation between latent motivation and demographic variables Social norms Informed on Demographic variable Pro Env 1 Pro-Social Pro Env 2 Pro-Govt (std) PES (std) Household size -0.04 0.01 0.06 -0.11 0 -0.09 Gender of household head (1=male) 0.02 0.1 -0.04 0 -0.1 -0.05 Age of household head -0.07 0.08 -0.06 -0.02 -0.12 0.02 Education level of household head 0.01 -0.06 0.13 0.11 0.05 0 Household income from agriculture 0.03 -0.07 0.03 0.03 0.08 0.09 If household uses credit -0.09 0.01 -0.1 0.07 0.08 0.09 Profit (in logs) -0.05 0.03 0.02 0.02 -0.03 0.03 Last yr income not typical (earned less) 0.02 0.05 -0.11 -0.07 0.05 -0.06 Area of property (in ha) -0.11 -0.01 -0.07 0.06 0.01 0.07 Number of agricultural workers 0 0.04 -0.09 -0.11 -0.05 0.06 Possess legal documents 0 0.05 -0.11 0.02 -0.01 0.09 Land type: Clay -0.05 0.08 0 0.07 0.02 -0.26 Land type: Sand 0.03 -0.02 0.1 0.01 -0.04 0.07 Land type: Clay-Sand -0.04 0.04 -0.06 -0.08 0.06 0.03 Land type: Terra Roxa -0.04 0.03 -0.11 0.03 -0.08 0.06 Proportion of steep land in total property -0.16 0.04 -0.11 -0.11 -0.02 -0.05 Number of properties eroded -0.02 0.06 -0.06 -0.02 -0.13 -0.06 Land used for agriculture 0.07 0.1 -0.09 0.01 -0.07 0.15 Area of APPs conserved with trees -0.05 0.05 0.03 0.05 -0.11 -0.01 Area of non APPs conserved with trees 0.15 0.03 0.1 -0.06 0.02 -0.07 12 Two non-factored indices were constructed by adding dichotomous variables and then standardizing the index with a mean of 0 and standard deviation of 1: Social norms The respondent plans to discuss the project with neighbors after the survey. The respondent would participate in the program if they found out neighbors are participating. Informed Heard of PES or a similar scheme before. Heard of the Mina d’Agua project before. Fully understands the Forest Code. We might be interested in how intrinsic motivations are related to observable characteristics. While observed baseline conservation and proxies for opportunity costs are crucial for understanding take up, these observables fall short in capturing all of the noise in the underlying drivers of take up. By also measuring intrinsic motivations, we aim to explain more of the mechanisms in addition to proxies for opportunity costs. Thus, we should observe very low correlation between our intrinsic motivations and the observable characteristics of the landholders, as seen in Table 3. To verify the indices are also independent of each other, Table 4 displays the correlation matrix of the indices measuring the latent motivations. Table 4: Correlation matrix of indices Social Informed norms on PES Pro Env 1 Pro-Social Pro Env 2 Pro-Govt (std) (std) Pro Env 1 1 Pro-Social -0.0744 1 Pro Env 2 0.1068 -0.0955 1 Pro-Govt 0.1175 -0.0478 0.2034 1 Social norms (std) 0.0926 -0.2174 0.0918 0.0907 1 Informed on PES (std) -0.0561 -0.0258 -0.0564 0.0093 0.0133 1 5. Estimation strategy As described earlier, we use a two-step method for our quantitative analysis. First we estimate if stated preferences, measured by factor analysis, predict revealed preferences of conservation behavior on land under legal protection (Model 1) and conservation behavior on land outside legal protection (Model 2). Then with validation that the indices predict conservation behavior, we use the indices to explain PES program take up when controlling for a set of observable proxies for opportunity costs (Model 3). Stated versus Revealed Preferences model First we test if the stated preferences as captured by the indices predict revealed preferences as reflected in the level of existing conservation on the property 13 at the time of the baseline survey. The hypothesis is that those with Pro-Environment and Pro-Social motivations will already be conserving land that is not under legal protection, controlling for observable proxies for opportunity costs as explained below. Conservation is specified as the amount of forest cover both within APPs, and as a separate outcome variable, on the remainder of the landholder’s property outside of APPs (Table A4 and A5 in Appendix). The starting point is strict homo economicus landholders. In line with the expectations of rational choice models of behavior, these landholders conserve only if the net benefit of doing so is positive, taking into account the opportunity cost of foregone revenue from alternative uses of the land. Proxies for opportunity cost controls are captured by the SocioEconomics’, and LandCharacteristics’i vectors. We define these as observable opportunity costs because from a landholder’s point of view, the quality of their land, inputs and income determine their willingness to accept the incentive for the program.8 Conservation beyond the expected level of a strict homo economicus landholder suggests the individual may hold attitudes that lead him or her to conserve some portion of land irrespective of costs and benefits. The remaining vectors capture these attitudes. Model 1: Percent APP Conservedi,t=0 = α + β1.SocioEconomics’i + β2.LandCharacteristics’i + β3.ProEnv’i + β4.ProSocial’i + β5.ProGovt’i + β6.SocialNorms’i + β7.Informed’i + εi and Model 2: Percent NonAPP Conservedi,t=0 = α + β1.SocioEconomics’i + β2.LandCharacteristics’i + β3.ProEnv’i + β4.ProSocial’i + β5.ProGovt’i + β6.SocialNorms’i + β7.Informed’i + εi Where i is household at time=0 (baseline survey). As the outcome variable in both equations is a proportion, we used a generalized linear model (GLM) with a logit link and the binomial family. Standard errors are clustered at the municipality level, as t- tests indicate significant differences in means (see Appendix). Standard errors are bootstrapped with 1000 replications. SocioEconomics’i is a vector of covariates conventionally used in PES take up analysis and include education, age, and gender of the head landholder, total income and (log) profits from agriculture,9 credit access and use, and if last year was typical 8 Income may correlate with unobserved factors, such risk and time preference as well as intellectual ability and competence. If this were the case, then the variable income would capture the joint effect of opportunity cost and these unobserved factors on conservation and program participation decisions. 9 Parameter tests and likelihood ratio tests lead us to restrict our model to two proxies for income (income shock last year and credit use). 14 (and if not, if the landholder earned less or more income). These variables are critical for assessing opportunity cost; if the landholder has significant profits from agriculture, they may be less likely to enroll in a PES program unless the payments for the program are higher than their profits from land use. Information on credit access and use are helpful to understand if people have the ability or willingness to take investments on their land. Alternatively, if they are credit constrained they may need a conditional cash transfer program to overcome investments needed for conservation/restoration. LandCharacteristics’i is a vector of covariates including the size of property in hectares, number of people working on the land, possession of required legal documents for ownership or renting of the property, soil characteristics (sand, clay, mix, red soil), steepness of land (proportion of property with steep parcels), evidence of erosion on property, number of springs on property, and if the property is used for agriculture. Included are also variables regarding the landholder’s plans to deforest their land or the trees around the spring. Although it is an empirical question, if the landholder has more land, and more workers on the land, we may expect them to be more willing diversify land use and to enroll part of their property in a PES program compared to a smaller landholder who may not have the ability to diversify their land use. As this depends on the motivations of the landholder, we will observe the effects when including motivations as independent variables (see below). Those with property rights that are well defined may be more likely to invest and take care of their land. If the land suffers from high erosion, the landholder may be more willing to participate in a restoration program. Furthermore, flatter land is expected to be more productive so conserving it would entail higher opportunity costs. Pro-Environmenti, Pro-Sociali, Pro-Govti, and Social Normsi, and Informedi are as defined in the previous section. Participation model We estimate landholder-level probit regressions (Table A6 in Appendix) where the dependent variable Yi is a dichotomous variable equal to 1 if landholder i enrolled in the PES program, and 0 if not. Municipality fixed effects are included. Marginal effects are computed for continuous and dichotomous explanatory variables. Model 3: Participatei = α + β1.SocioEconomics’i + β2.LandCharacteristics’i + β3.ProEnv’i + β4.ProSocial’i + β5.ProGovt’i + β6.SocialNorms’i + β7.Informed’i + εi Vectors SocioEconomic’, LandCharacteristics’, Pro-Env’, Pro-Social’, ProGovt’, SocialNorms’, and Informed’ remain. 15 4. Results Stated versus Revealed Preferences (Models 1 and 2) Opportunity costs Observable opportunity costs are predictive of conservation behavior on land under legal protection (within APPs). As the total area of property in hectares increases, conservation of APPs increases by 2 percent (Table A4). Landholders have more land to divert their land use activities outside of the APPs. Further, as the total number of properties eroded increases by one, conservation of APPs increases by 27 percent. As we might expect, opportunity costs are more strongly predictive of conservation behavior on land outside of legal protection. If the land is mostly used for agriculture, conservation outside APPs decreases by 68 percent (Table A5). As the number of agricultural workers on the property increase by one, conservation of land outside of APPs decreases by 10 percent. We see strong negative associations when analyzing income dynamics: conservation outside APPs decreases by 25-27 percent if the landholder experienced an income shock in the year prior and earned less income than usual. However, forest cover changes slowly. One would not expect forest cover to respond to short-term fluctuations, so the size of this impact in correlations is surprising. Motivations In pairwise correlations, the percent of APP conserved is not highly correlated with the percent of non-APP area conserved (12 percent). Thus we may assume different motivations are at play for choosing to conserve land under legal protection versus land outside of legal protection. Pro-Social landholders—those who are concerned about protecting the environment for future generations—are more likely to conserve land both within and outside legal protection. In both regressions, when controlling for proxies for opportunity costs as defined by socioeconomic and land use vectors, the coefficient is positive and statistically significant: a one standard deviation increase in the index results in a 7 percent increase in conservation within APPs, and a 10 percent increase in conservation outside APPs. Therefore these findings are in line with the hypothesis that Pro-Social motivations drive conservation both inside and outside APPs. Pro-Environment landholders- those who are concerned about the future supply of water and those concerned about the negative impact of land degrading activities on the environment - are more likely to conserve land outside APPs (Table A5). Controlling for opportunity costs, if the landholder believes the future supply of water will diminish, conservation of land outside of APPs increases by 80%. A one unit standard deviation increase in the concern of negative land use activities results in a 7 percent increase in conservation outside APPs. This is in line with the hypothesis that those who are concerned for the environment would conserve irrespective of the zoning laws as they value the existence of the environment. The results are robust under different specifications. 16 These findings are in line with the hypotheses that Pro-Environment and Pro- Social motivations drive conservation both within and outside APPs. The results also confirm the hypothesis that stated preferences of conservation are strong predictors of revealed preferences in our survey. This result provides a robustness check for using the factored indices to proxy latent motivations as independent variables in the take up analysis. However, if the landholder believes that restoration of border areas of rivers and springs through public resources would significantly improve air quality and the environment, conservation of land outside APPs decreases by 12%. Recall that this motivation was perfectly correlated with attitudes towards government’s role in protecting natural resources from Factor Analysis 3 (A3 in the Appendix): those who acknowledge benefits of restoration yet believe it’s the government’s responsibility to protect and pay to protect natural resources on public property are not incentivized to conserve in absence of financial support. Take Up of PES Program (Model 3) Opportunity costs Opportunity costs are strong predictors of PES enrollment. When participation is regressed only on observable opportunity costs, we see that as household size increases by one member, landholders are 2 percent less likely to participate (Table A6 in the Appendix). Landholders who suffered an income shock and earned less income than usual in the previous year are 10 percent less likely to participate. As the proportion of very steep property increases, the landholder is 15 percent more likely to enroll in the program. Furthermore, landholders who have all required legal documents for their property are 35 percent more likely to enroll in the program. The proxy variables for opportunity cost remain significant and at similar, if not slightly smaller, magnitudes in the fully specified model. We asked if additional incentives, such as access to technical assistance, would change their participation decision. Free mapping of the property, fencing, seeds/seedlings and specialized technical assistance were provided as options. An overwhelming 92 percent of landholders replied that access to assistance in any form would not change their participation decision. Motivations Those with Pro-Social motivations are less likely to enroll in the PES program. A one-unit standard deviation increase in Pro-Social motivation leads to a 5 percent decrease in probability of participating in the program in the fully interacted model. The finding that Pro-Social motivated landholders are more likely to be conserving land both within and outside APPs before the program is introduced, yet are less likely to then enroll in the program, demonstrates a potential crowding out effect of a government incentives program on Pro-Social motivated landholders. However, we can only evaluate if crowding out of conservation behavior occurred once the program has finished. 17 Those concerned about social norms also are less likely to enroll in the PES program. A one-unit increase in standard deviation in Social norms also has a negative effect on participation; the landholder is 7.8 percent less likely to participate. And if the social norm motivated landholder received a high offer, they are 22 percent less likely to enroll. Recall the social norms indicator is comprised of responses to two questions: if the landholder plans to discuss the project with their neighbor after the survey, and if the landholder would change their decision based on the decision of their neighbor. Our original hypothesis holds that the landholder’s participation decision is dependent on their valuation of how conservation behavior, and how accepting a monetary reward in return for positive conservation outcomes, is judged by their community. The social connection moderator is triggered as approaching each landholder individually undermines the social connections within the community. Pro-Environment motivated landholders - on the other hand - are 3 percent more likely to enroll in the program. The incentives program may be activating all psychological moderators under SDT theory: the incentive may be supporting their self-determination and competence if landholders are acting in accordance with their free will to conserve (as shown in the baseline conservation regressions) and value the compensation as positive feedback on their existing conservation behavior. It also may support their environmental connection if it further supports their desire to feel connected to nature. Most importantly, access to information is the strongest predictor of take up. Landholders who had heard of PES in general and/or of the specific MdA program, and who fully understand the Forest Code are 13.2 percent more likely to participate in the program. 4. Qualitative Follow-up Survey A qualitative survey was administered to those eligible for the program to understand why they accepted or rejected the incentive to enroll in the conservation program. Please note that we do not include these questions in our quantitative analysis as they were administered to the relative subsamples (enrolled versus non enrolled) and thus the quantitative analysis would suffer from selection bias. Furthermore due to the small sample, we would not have enough power to quantitatively estimate the patterns in responses. Instead we highlight the results from the qualitative survey below, which serve as further support of our hypothesized channels through which an incentive supports or undermines motivation and motivations for enrollment. When asked specifically why the household chose to participate in the program the most common response was “the payment is little but it is an incentive to comply with the law” (Figure 5). This confirms our quantitative finding that those with access to information are 15 percent more likely to enroll in the program. In line with SDT, the payments may enhance self-determination for those well informed through the channel of self-development by learning the legal requirements and context. 18 Would restore anyway The pay is good Payment is small, but it is an incentive to comply with the law The payment is greater than the expenses I will have I need the money 0 10 20 30 40 50 60 % of households Figure 5: Motives for enrolling in the PES restoration program According to our qualitative findings, just over one-quarter of those who accepted the offer were planning to restore regardless of the program. As mentioned earlier, many hypothesize that low rates of reduced deforestation in PES schemes are due to low additionality whereby those already conserving self select into the program. Here we qualitatively observe a possible additionality issue. Among the 27 percent of households who indicated that they would restore anyway, the most common reason for doing so was to conserve water they consume (Figure 6). Here we observe validation from our quantitative findings that those who are Pro-Environment are more likely to conserve land before the introduction of the program, and are more likely to also enroll in the PES program. The area is not productive To meet a legal obligation To conserve water that I consume To conserve water that others consume Other 0 5 10 15 20 25 30 35 % of households intending to restore anyyway Figure 6: Motives for intending to restore anyway 19 Questions were also asked to those who chose not to enroll. The most common response was “the incentive was low” for the conservation program and “recovering a hectare compromises a very productive area” (Figure 7). I should not get paid for it as its my obligation Compromises very productive area The payment is too low Not confident the program will be implemented It is not my responsibility Conservation Would use the area for other purposes Restoration I do not need the money I do not want someone to control what I do on my property Bureaucracy I do not have the necessary documents I don't know/Other 0 5 10 15 20 25 30 35 % of households that didn't enroll Figure 7: Motives for not enrolling in the PES program Thus for both the conservation and restoration program, opportunity costs are a crucial hurdle to increase enrollment. Respondents also mentioned that they did not want someone to control their behavior on their own property. This response suggests an undermining of self-determination by the principal. According to SDT theory, the incentive may feel imposed on the individual and result in a crowding out of their motivation. 5. Conclusion Standard take up analyses of PES programs study the impact of observable opportunity costs on enrollment. We extend this work by further incorporating behavioral determinants (intrinsic motivations) in the enrollment decision. We first disentangle and measure intrinsic motivations, specifically Pro-Environment, Pro-Social, Pro- Government, and Social Norms. Controlling for proxies for opportunity costs, we then analyze behavioral determinants of take up for the PES program. We discuss the findings in light of SDT, which hypothesizes mechanisms through which payments may alter the landholder’s self-satisfaction need to feel competent, self-determined, and connected to others and their environment (“social relatedness” and “environmental relatedness”). The specific design of payments affects these need satisfaction moderators in different ways. The psychological moderators, individually or in 20 combination, drive the crowding in or crowding out effect of a PES incentive on take up. To first measure latent motivations, we use factor analysis on responses to an exhaustive and multi dimensional questionnaire in the baseline survey for a PES program in the state of São Paulo, Brazil. We then determine if stated conservation preferences predict revealed conservation preferences, as determined by existing level of forest cover on their property before the program is introduced. The program conditions require conservation only of land already under legal protection by the Brazil Forest Code. We find that Pro-Social landholders are more likely to be conserving land both within and outside areas of legal protection (APPs) before the program is introduced. These findings are in line with the hypothesis that Pro-Social motivations drive conservation both within and outside APPs. Pro- Environment motivations drive conservation on land outside APPs. The results also confirm the hypothesis that stated preferences of conservation are strong predictors of revealed preferences in our survey. This result provides a robustness check for using the factored indices to proxy latent motivations as independent variables in the take up analysis. We use the indices to then study determinants of take up to analyze which types of individuals are more likely to enroll in the PES program when controlling for proxies for opportunity costs. Landholders with a low opportunity cost of conservation are already conserving and, in turn, are more likely to enroll in a program which compensates them for their existing behavior. Thus we would then expect that those with high opportunity costs would not enroll in the program. This finding is supported by the qualitative follow-up survey: the main reason provided to not enroll in the restoration program was because it compromises a productive area of land. When analyzing behavioral determinants of take up, we observe Pro- Environment landholders are more likely to enroll in the PES program. The payments may support the landholder’s need to feel connected to their environment, and the incentive works to crowd in their motivation. This finding is reinforced by the qualitative follow up survey: the main reason landholders enrolled was to protect water resources. The crux of SDT theory is that self-determination support is necessary to maintain intrinsic motivation. Payments may be seen as a type of positive feedback on a person’s performance (competence), which can increase intrinsic motivation (Deci, 1971). It is worth noting these landholders are closest to the margin of adopting - as they were already conserving land outside of legal protection but not yet conserving land under legal protection - and a small incentive crowds in their intrinsic motivation. While we found that Pro-Social landholders, like Pro-Environment landholders, are more likely to be conserving land regardless of the legal requirements before the program is introduced, these landholders are less likely to enroll in the program. Drawing upon SDT theory, the incentive may undermine their free will and desire to feel connected to their community through their Pro-Social behavior. Under this model, such landholders may even reduce their conservation. Further research should 21 explore the dynamic effects of whether the introduction of a monetary incentive program for conservation results in a crowding out of observed conservation behavior in the long run. Landholders with social norms motivations are also less likely to enroll in the program. As the program was introduced to each landholder individually, this may have undermined the social connection moderator of the landholder’s needs satisfaction. Importantly, one of the largest drivers of enrollment is having information on the Forest Code, the concept of PES, and information on this specific PES program before the survey was administered. This is a key finding for program administrators. Education campaigns that provide this information are likely to increase enrollment. While individual motivations must be taken into account to assure that PES is cost- effective and results in high additionality, we find the opportunity cost of land is still the most critical hurdle to change land use. If program administrators are concerned that those already engaging in conservation practices self-select into the program, then it is advisable to direct the monetary incentive offer to landholders not conserving because of high opportunity costs to motivate a change in existing behavior and ensure additional conservation. The survey responses suggest that other incentives such as technical assistance of any kind would not change their participation decision, nor would additional financing. Further research is needed to understand the most effective combination of incentives to promote conservation for those with high opportunity costs of land use. Our paper sheds light on the importance of using preference questions in a baseline survey to analyze pre-existing motivations before the introduction of a payments program for conservation. These questions can be utilized to gain a more comprehensive understanding of the reasons landholders chose whether to enroll in the program. Understanding how pre-existing motivations interact with the decision to participate in a PES program can bring important insight to contract design under asymmetric information. If program administrators want to achieve high additional conservation, they could use this information to more efficiently target programs to those who are not yet conserving their land according to program conditions. Further, by repeating studies to have a firmer understanding of the effect of a monetary offer on various intrinsic motivations, program administrators can use the analysis to direct payments to those close to the margin of conserving but need a incentive to comply with the program conditions (like the Pro-Environment landholders in our sample). They should not offer a monetary incentive to those whose pre-existing intrinsic motivations are undermined by the extrinsic incentive (like the Pro-Social landholders in our sample). To the extent that people do have pre-exiting motivations to conserve, implementation of a PES program might crowd out their conservation behavior in the long run, whereas if no such motivations existed previously, or the motivations have no effect on participation, then crowding out is unlikely to be an issue. However, further research is needed to understand the long run effect of the introduction and implementation of PES in the community on conservation behavior for both those who did and did not enroll in the program. 22 References Alix-Garcia, J.M., K.R.E. Sims, and P. Yañez-Pagans. 2015. “Only one tree from each seed? Environmental effectiveness and poverty alleviation in Mexico's Payments for Ecosystem Services Program.” American Economic Journal: Economic Policy, 7(4):1-40. Alix-Garcia, J.M., E.N. Shapiro, and K.R.E. Sims. 2012. “Forest conservation and slippage: Evidence from Mexico’s national payments for ecosystem services program.” Land Economics, 88(4):613-638. Arriagada, R.A., P.J. Ferraro, E.O. Sills, S.K. Pattanayak, and S. Cordero-Sacho. 2012. “Do payments for environmental services affect forest cover? A farm-level evaluation from Costa Rica.” Land Economics, 88(2):382-399. Fiszbein, A., N. Schady, F.H.G. Ferreira, M. Grosh, N. Keleher, P. Olinto, and E. Skoufias. 2009. Conditional Cash Transfers: Reducing Present and Future Poverty. Washington: World Bank. Carrascosa von Glehn, H.Q., D. Taffarello, A. Kamiyama, A.C. Dalla Vecchia, and C. Vigo Cogueto. 2012. “Projeto Mina d’Água.” In: S. Pagiola, H. Carrascosa von Glehn, and D. Taffarello (Eds.), Experiências de pagamentos por serviços ambientais no Brasil. São Paulo: Secretaria do Meio Ambiente. Clark, C.F, M.J. Kotchen, and M.R. Moore. 2003. “Internal and external influences on pro- environmental behavior: Participation in a green electricity program.” Journal of Environmental Psychology, 23(3):237-246. Deci, E.L. 1971. “Effects of externally mediated rewards on intrinsic motivation.” Journal of Personality and Social Psychology, 18(1):105–115. Deci, E.L., and R.M. Ryan. 1985. “The general causality orientations scale: Self-determination in personality.” Journal of Research in Personality, 19(2):109-134. Deci, E.L., and R.M. Ryan. 1991. “A motivational approach to self: Integration in personality.” In: R.A. Dienstbier (Ed.), Perspectives on Motivation. Lincoln: University of Nebraska Press. Engel, S., S. Pagiola, and S. Wunder. 2008. “Designing payments for environmental services in theory and practice: An overview of the issues.” Ecological Economics, 65(4):663–674. Ezzine-de-Blas, D., E. Corbera, and R. Lapeyre. 2015 “Crowding-in or crowding-out? A conceptual framework to understand motivations in payments for ecosystem services.” Brighton: Institute of Development Studies. Ferraro, P.J., and A. Kiss. 2002. “Direct payments to conserve biodiversity.” Science, 298(5599):1718–1719. Ferraro, P.J., and S.K. Pattanayak. 2006. “Money for nothing? A call for empirical evaluation of biodiversity conservation investments.” PLoS Biology, 4(4):482-488. Frey, B. 1992. “Tertium datur: Pricing, regulating and intrinsic motivation.” Kyklos, 45(2):161–184. Frey, B., and F. Oberholzer-Gee. F. 1997. “The cost of price incentives: An empirical analysis of motivation crowding-out.” The American Economic Review, 87(4):746–755. Gertler, P. 2004. “Do conditional cash transfers improve child health? Evidence from PROGRESA's control randomized experiment.” American Economic Review, 94(2):336-341. McCauley, D.J. 2006. “Selling out on nature.” Nature, 443:27-28. 23 Nolan, J.M., P.W. Schultz, R.B. Cialdini, N.J. Goldstein, and V. Griskevicius. 2008. "Normative social influence is underdetected." Personality and Social Psychology Bulletin, 34(7):913- 923. Padovezi, A., R.A. Gorne Viani, U. Kubota, D. Taffarello, M. Faria, H. Bracale, V. Ferrari, and F.H. de Carvalho. 2012. “Produtor de Água na bacia hidrográfica PCJ.” In: S. Pagiola, H. Carrascosa von Glehn, and D. Taffarello (Eds.), Experiências de pagamentos por serviços ambientais no Brasil. São Paulo: Secretaria do Meio Ambiente. Pagiola, S. 2012. “Using PES to implement REDD.” PES Learning Paper No.2011-1. Washington: World Bank. Pagiola, S., and G. Platais. 2007. Payments for Environmental Services: From Theory to Practice. Washington: World Bank. Pagiola, S., and A. Rios. 2013. “Evaluation of the impact of payments for environmental services on land use change in Quindío, Colombia.” PES Learning Paper No.2013-2. Washington: World Bank. Pagiola, S., A.R. Rios, and A. Arcenas. 2008. “Can the poor participate in payments for environmental services? Lessons from the Silvopastoral Project in Nicaragua.” Environment and Development Economics, 13(3):299–325. Pagiola, S., J. Honey-Rosés, and J. Freire-González. 2016. “Evaluation of the permanence of land use change induced by payments for environmental services in Quindío, Colombia.” PLOS One, 11(3):e0147829. Pattanayak, S.K., S. Wunder, and P.J. Ferraro. 2010. “Show me the money: Do payments supply environmental services in developing countries?” Review of Environmental Economics and Policy, 4(2):254-274. Robalino, J.A., and A.S.P. Pfaff. 2013. “Ecopayments and deforestation in Costa Rica: A nationwide analysis of PSA’s initial years.” Land Economics, 89(3):432-448. Ryan, M.R., and E.L. Deci. 2000. “Intrinsic and extrinsic motivations: Classic definitions and new directions.” Contemporary Educational Psychology, 25:54-67. Sierra, R., and E. Russman. 2006. “On the efficiency of environmental service payments: A forest conservation assessment in the Osa Peninsula, Costa Rica.” Ecological Economics, 59(1):131–141. Sills, E., R. Arriagada, P. Ferraro, S. Pattanayak, L. Carrasco, E. Ortiz, S. Cordero, K. Caldwell, and K. Andam. 2008. “Impact of Costa Rica’s Program of Payments for Environmental Services on land use.” PES Learning Paper No.2008-3. Washington: World Bank. Skoufias, E., Parker, S.W., Behrman, J.R. and Pessino, C., 2001. “Conditional cash transfers and their impact on child work and schooling: Evidence from the PROGRESA program in Mexico [with comments].” Economia, 2(1):45-96. Wunder, S. 2005. “Payments for environmental services: Some nuts and bolts.” Occasional Paper No.42. Bogor: CIFOR. Wunder, S., S. Engel, and S. Pagiola. 2008. “Taking stock: A comparative analysis of payments for environmental services programs in developed and developing countries.” Ecological Economics, 65(4):834–852. 24 Appendix Photos: Stefano Pagiola Figure A1: Examples of well-conserved springs in Guapiara municipality, with extensive vegetation cover around the spring (the legal requirement is for vegetation in a 50m radius around the spring to be conserved) 25 Photo: Stefano Pagiola Figure A2: Example of poorly conserved spring in Guapiara municipality, with minimal vegetation cover around the spring and steep cultivated slopes in the recharge area above 26 Photos: Stefano Pagiola Figure A3: Examples of farming landscapes in Guapiara municipality, with cultivated areas and pastures, often on steep slopes, as well as forest remnants and eucalyptus plantations 27 Table A1: Characteristics of landholders in the study areas Total Enrolled Non-enrolled Ibiúna Guapiara Mean s.d. Mean s.d Mean s.d. Mean s.d. Mean s.d. Household size 3.92 2.68 3.48 1.7 3.73 2.35 4.05 2.22 3.76 3.13 Gender of household head (1=male) 0.85 0.35 0.86 0.35 0.81 0.4 0.88 0.33 0.83 0.38 Age of household head 55.37 13.89 55.45 13.8 54.34 13.27 54.05* 14.63 56.9 12.85 Education level of household head 2.47 1.16 2.6 1.19 2.41 1.15 2.81*** 1.25 2.07 0.9 Household income from agriculture 1.35 1.64 1.19 1.44 1.54 1.82 1.28 1.71 1.44 1.56 If household uses credit 0.15 0.35 0.12 0.33 0.17 0.38 0.11** 0.31 0.19 0.39 Profit (in logs) 12.13 0.14 12.13 0.1 12.13 0.13 12.12 0.18 12.14 0.08 Last year’s income lower than usual 0.29 0.45 0.21** 0.41 0.34 0.47 0.16*** 0.37 0.44 0.5 Area of property (in ha) 12.24 13.07 13.01 14.3 11.04 13.07 12.31 13.06 12.15 13.13 Number of ag workers 1.97 2.17 1.68 1.45 1.91 1.82 1.52*** 1.94 2.49 2.31 Possess legal documents 0.88 0.33 0.95*** 0.21 0.83 0.38 0.85 0.36 0.91 0.29 Land type: Clay 0.26 0.44 0.17 0.38 0.17 0.37 0.32*** 0.47 0.19 0.39 Land type: Sand 0.26 0.44 0.32 0.47 0.27 0.45 0.34*** 0.47 0.17 0.37 Land type: Clay-Sand 0.23 0.42 0.25 0.44 0.2 0.4 0.26 0.44 0.19 0.39 Land type: Terra Roxa 0.44 0.5 0.37** 0.49 0.54 0.5 0.26*** 0.44 0.64 0.48 Proportion of property with steep land 0.76 0.28 0.77 0.28 0.72 0.29 0.77 0.26 0.74 0.3 Number of ha eroded 0.15 0.49 0.11 0.37 0.15 0.44 0.06*** 0.27 0.25 0.65 Uses land for agriculture 0.8 0.4 0.79 0.41 0.84 0.37 0.7*** 0.46 0.93 0.26 Plans to deforest spring 0.04 0.2 0.08* 0.27 0.02 0.14 0.03 0.18 0.06 0.23 Conserve land in APPs 0.75 0.33 0.77 0.33 0.71 0.37 0.76 0.33 0.74 0.33 Conserve land in Non APPs 0.37 0.33 0.38 0.32 0.33 0.33 0.42*** 0.32 0.31 0.32 Interested to Participate 1.68 0.86 1.59 0.81 1.67 0.88 1.58** 0.81 1.8 0.91 Notes: * p<0.10, ** p<0.05, *** p<0.01 T-tests of means: EOI vs. Non-EOI; Guapiara vs. Ibiuna 28 Table A2: Factor Analysis II: Community and landholders’ responsibility to protect and pay (rotated factors) Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Responsible for protecting public resources: Community or All -0.08 0.11 -0.01 0.66 -0.03 -0.02 0.17 Community only -0.04 -0.06 -0.03 0.66 0.11 0.03 -0.05 End users 0.10 -0.03 -0.02 -0.06 -0.01 0.00 0.04 Responsible for paying to protect public resources: Community or All -0.07 0.13 0.01 0.13 0.22 0.13 0.57 Community only 0.02 0.01 -0.02 0.17 0.53 -0.04 0.28 End users 0.04 0.10 -0.02 0.02 -0.02 -0.04 -0.10 Responsible for protecting private resources: Landowners -0.02 -0.71 0.02 0.01 -0.17 -0.06 -0.03 Community or All 0.03 0.78 -0.01 0.04 -0.09 0.04 0.04 End users -0.01 0.10 0.00 -0.06 0.46 -0.02 0.05 Community only -0.02 0.36 0.00 -0.03 -0.03 -0.08 -0.07 Responsible for paying to protect private resources: Landowners 0.12 -0.10 0.06 -0.05 -0.01 -0.14 0.10 Community or All -0.06 0.24 -0.02 -0.03 -0.04 0.55 0.14 End users -0.02 -0.01 -0.01 0.05 -0.08 -0.04 0.23 Community only -0.02 -0.03 -0.01 0.08 -0.01 0.48 0.00 Responsible for protecting the environment for future generations: Community, All or Landowners 0.08 0.05 0.03 0.00 0.06 -0.05 0.01 Landowners only 0.99 0.01 -0.13 -0.02 0.00 -0.01 -0.01 Community only 0.10 -0.01 0.99 -0.01 0.00 0.00 0.00 Community or Landowners 0.95 0.01 0.32 -0.02 0.00 -0.01 -0.01 Eigenvalue 1.94 1.37 1.12 0.94 0.61 0.59 0.54 Proportion of Variance 0.31 0.22 0.18 0.15 0.10 0.09 0.09 Notes: Shaded cells show within each factor, those with high factor loadings in the same direction, and thus included in an index together. 29 Table A3: Factor Analysis III: Government’s responsibility to protect and pay (rotated factors) Factor 1 Factor 2 Factor 3 Pro-Government: Responses to survey questions Pro-Govt - - Responsible for protecting water resources on public land: Government only: Government only 0.039 0.364 -0.008 Responsible for paying to protect water resources on public land: Government only 0.376 0.662 -0.000 Responsible for protecting water resources on private land: Government only 0.019 -0.017 0.039 Responsible for paying to protect water resources on private land: Government only 0.882 -0.016 -0.003 Responsible for protecting water resources on private and public land: Government only 0.883 0.335 0.003 Eigenvalue 1.701 0.684 0.002 Proportion of Variance 0.776 0.312 0.001 Notes: Shaded cells show within each factor, those with high factor loadings in the same direction, and thus included in an index together. 30 Table A4: Stated versus revealed conservation in APPs (Model 1) I II III Means Conventional Indices Household size 3.928 -0.022 -0.03 (0.145) (0.173) (0.174) Education level of household head 2.474 -0.064* -0.067** (0.063) (0.033) (0.034) If household uses credit 0.145 -0.404 -0.483 (0.019) (0.324) (0.377) Last yr income not typical (earned less) 0.295 0.039 0.092 (0.025) (0.439) (0.503) Area of property (in ha) 12.296 0.022*** 0.021*** (0.706) (0.002) (0.002) Number of ag workers 1.977 -0.007 0.012 (0.117) (0.159) (0.161) Possess legal documents 0.879 -0.163 -0.16 (0.018) (0.137) (0.215) Land type: Sand 0.266 0.331* 0.289* (0.024) (0.197) (0.161) Land type: Clay 0.257 0.068 0.085 (0.024) (0.195) (0.216) Land type: Sand-clay 0.228 0.36 0.345 (0.023) (0.233) (0.256) Land type: Terra Roxa 0.434 0.051 0.032 (0.027) (0.033) (0.137) Proportion of total steepness over total property 0.756 0.576 0.535 (0.015) (0.422) (0.483) Number of properties eroded 0.145 0.274*** 0.260*** (0.026) (0.054) (0.058) Land used for agriculture 0.806 -0.306 -0.266 (0.021) (0.646) (0.791) Area of non-APP conserved with trees 0.373 0.529 0.635 (0.018) (0.6) (0.647) Concerned about future supply of water 0.234 -0.248 (0.023) (0.154) Concerned about protecting environment for future generations 0.004 0.074*** (0.054) (0.013) Concerned about behavioral impacts on environment 0.005 0.065 (0.054) (0.055) Pro public-financed restoration 0.002 0.129 (0.054) (0.136) Constant 0.703 0.759 (0.668) (0.558) Observations 346 341 337 Squared correlation between observed and predicted 0.07 0.08 Notes: Standard errors shown in parentheses; *p<0.10, **p<0.05, ***p<0.01 31 Table A5: Stated versus revealed conservation in non APPs (Model 2) I II III Means Conventional Indices Household size 3.928 0.080*** 0.079*** (0.145) (0.015) (0.012) Education level of household head 2.474 -0.073 -0.083 (0.063) (0.065) (0.075) If household uses credit 0.145 -0.163 -0.028 (0.019) (0.368) (0.273) Last yr income not typical (earned less) 0.295 -0.254*** -0.309*** (0.025) (0.064) (0.018) Area of property (in ha) 12.296 0 0.003 (0.706) (0.009) (0.009) Number of ag workers 1.977 -0.092** -0.101*** (0.117) (0.046) (0.033) Possess legal documents 0.879 -0.025 -0.027 (0.018) (0.066) (0.031) Land type: Sand 0.266 -0.197** -0.162*** (0.024) (0.083) (0.05) Land type: Clay 0.257 -0.057 -0.118*** (0.024) (0.073) (0.025) Land type: Sand-clay 0.228 -0.032 -0.025 (0.023) (0.134) (0.113) Land type: Terra Roxa 0.434 -0.22 -0.163 (0.027) (0.194) (0.186) Proportion of total steepness over total property 0.756 0.213 0.367 (0.015) (0.219) (0.26) Number of properties eroded 0.145 -0.444* -0.573*** (0.026) (0.269) (0.213) Land used for agriculture 0.806 -0.680** -0.788*** (0.021) (0.266) (0.221) Area of APPs conserved with trees 0.75 0.409 0.451 (0.331) (0.505) (0.522) Concerned about future supply of water 0.234 0.729*** (0.023) (0.088) Concerned about protecting env for future generations 0.004 0.096** (0.054) (0.041) Concerned about behavioral impacts on env 0.005 0.067*** (0.054) (0.018) Pro Public financed restoration 0.002 -0.122*** (0.054) (0.033) Constant -0.055 -0.303 (0.325 (0.343) Observations 346 341 337 Squared correlation between observed and predicted 0.14 0.20 Notes: Standard errors shown in parentheses; *p<0.10, **p<0.05, ***p<0.01 32 Table A6: Program take up (Model 3) Enrolled in MdA I II III IV Means Conventional Indices Interaction effects b/se b/se b/se b/se Ibiúna 0.538 0.02 0.078*** 0.059*** (0.027) (0.031) (0.015) (0.01) Household size 3.928 -0.018*** -0.018*** -0.016*** (0.145) (0.003) (0.003) (0.005) Education level of household head 2.474 0.006 -0.001 0.041*** (0.063) (0.032) (0.006) (0.007) If household uses credit 0.145 -0.067 -0.093** -0.094 (0.019) (0.047) (0.047) (0.061) Last yr income not typical (earned less) 0.295 -0.099*** -0.048 -0.023 (0.025) (0.019) (0.033) (0.044) Area of property (in ha) 12.296 0.001 0.001 0.001 (0.706) (0.002) (0.003) (0.003) Number of ag workers 1.977 0.006 -0.002 -0.005 (0.117) (0.019) (0.012) (0.007) Possess legal documents 0.879 0.345** 0.317*** 0.270*** (0.018) (0.146) (0.056) (0.025) Land type: Sand 0.266 -0.162*** -0.096*** -0.071* (0.024) (0.04) (0.037) (0.037) Land type: Clay 0.257 -0.116 -0.084 -0.084 (0.024) (0.216) (0.2) (0.187) Land type: Sand-clay 0.228 -0.067*** -0.022 -0.040* (0.023) (0.019) (0.053) (0.022) Land type: Terra Roxa 0.434 -0.25 -0.189 -0.196* (0.027) (0.154) (0.13) (0.112) Proportion of total steepness over total property 0.756 0.152** 0.127*** 0.133*** (0.015) (0.063) (0.013) (0.013) Number of properties eroded 0.145 -0.047 -0.041 -0.012 (0.026) (0.12) (0.044) (0.059) Land used for agriculture 0.806 0.032 -0.001 0.044 (0.021) (0.046) (0.119) (0.122) 33 Table A6: Program take up (Model 3) Enrolled in MdA I II III IV Means Conventional Indices Interaction effects b/se b/se b/se b/se Area of APPs conserved with trees 0.374 0.082 0.044 0.046 (0.328) (0.148) (0.167) (0.151) Concerned about future supply of water -0.033 -0.039 (0.123) (0.144) Concerned about protecting env for future generations -0.021 -0.048** (0.013) (0.022) Concerned about behavioral impacts on env 0.044*** 0.042* (0.009) (0.025) Pro public financed restoration -0.022 0.008 (0.052) (0.043) Social norms -0.059** -0.033*** (0.029) (0.001) Access to info 0.129*** 0.145*** (0.003) (0.004) High offer 0.483 0.074 0.084 (0.027) (0.094) (0.087) Concerned about future supply of water * high offer 0.103 (0.088) Concerned about protecting env for future generations * high offer 0.187 (0.116) Concerned about behavioral impacts on env * high offer 0.021 (0.214) Pro public financed restoration * high offer -0.157 (0.12) Social norms * high offer -0.224*** (0.002) Access to info * high offer -0.086 (0.112) Observations 346 204 201 201 Squared correlation between observed and predicted 0.11 0.19 0.23 Notes: Standard errors shown in parentheses; *p<0.10, **p<0.05, ***p<0.01 34