Exploring the connections between participation in and benefits from payments for hydrological services programs in Veracruz State, Mexico

Exploring the connections between participation in and benefits from payments for hydrological services programs in Veracruz State, Mexico

Ecosystem Services 35 (2019) 32–42 Contents lists available at ScienceDirect Ecosystem Services journal homepage: www.elsevier.com/locate/ecoser Ex...

686KB Sizes 0 Downloads 48 Views

Ecosystem Services 35 (2019) 32–42

Contents lists available at ScienceDirect

Ecosystem Services journal homepage: www.elsevier.com/locate/ecoser

Exploring the connections between participation in and benefits from payments for hydrological services programs in Veracruz State, Mexico

T

Kelly W. Jonesa, , Sophie Avila Foucatg, Erin C. Pischkec, Jacob Salconea,d, David Torrezb, Theresa Selfae, Kathleen E. Halvorsenc,f ⁎

a

Human Dimensions of Natural Resources, Colorado State University, Ft. Collins, CO, USA División de Estudios de Posgrado, Facultad de Economía, Universidad Nacional Autónoma de México (UNAM), México City, Mexico c Department of Social Sciences, Environmental and Energy Policy Program, Michigan Technological University, Houghton, MI, USA d United Nations Environment Program, The Economics of Ecosystems and Biodiversity (TEEB), Geneva, Switzerland e Department of Environmental Studies, State University of New York-College of Environmental Science and Forestry, Syracuse, NY, USA f School of Forest Resources and Environmental Science, Michigan Technological University, Houghton, MI, USA g Instituto de Investigaciones Económicas, Universidad Nacional Autónoma de México, Mexico b

ARTICLE INFO

ABSTRACT

Keywords: Deforestation Ejido Environmental services Forest conservation Mexico Payments for ecosystem services

Payments for hydrological services (PHS) programs are expected to confer tangible benefits to households. Impact evaluations of PHS programs, however, find few to no changes in material indicators. One reason for this may be that non-financial motivations and benefits—including environmental or social—influence participation and are important outcomes for households participating in PHS programs. In this paper we test this hypothesis using 56 interviews and 181 surveys from households in Veracruz State, Mexico. Using logistic regression models we find that human, natural, physical and financial capital are important to the decision to participate in PHS, but so are pro-social and pro-environmental motivations. Using counterfactual impact evaluation methods we find few changes in material benefits but do find that PHS participants are more likely to report positive changes in their household and community quality of life over the last five years compared to households not participating in PHS programs. Qualitative information supports these findings. Our results contribute to the evolving theory on PHS that participation is driven by a mix of financial and non-financial motivations and that nonmaterial benefits are an important outcome of these programs.

1. Introduction Payments for hydrological services (PHS) programs are popular throughout Latin America as a strategy to address water quantity and quality concerns (Martin-Ortega et al., 2013; Bremer et al., 2016; Grima et al., 2016). They are part of a broader suite of tools known as payments for ecosystem services (PES). PES are voluntary transactions where beneficiaries of ecosystem services provide external incentives to ecosystem services providers, typically land owners, conditional on changes in land use or natural resource management behaviors (Wunder, 2015). The logic behind PES is that financial incentives will make conservation more attractive to land managers, inducing them to stop land uses that degrade ecosystem services (Wunder, 2006; Engel et al., 2008). The type of person that participates in PES programs directly influences the program’s ability to achieve its environmental or social objectives. Many PES programs target eligible areas based on expected



environmental outcomes. Within these eligible areas, participation is determined by both the ability to participate and the motivation to participate in the offered program (Pagiola et al., 2005). The ability to participate in PES programs is influenced by the opportunity costs of conservation relative to what is offered by the program and the ability of the individual to meet the investment needs of the PES program contract (Bottazzi et al., 2018). The latter includes the human, financial, physical, social, and natural capital assets available to the household. In many cases, the costs and barriers to entering PES programs have been found to be prohibitive for poorer farmers (Pagiola et al., 2005; Zbinden and Lee, 2005; Arriagada et al., 2009; Lansing, 2017). In general, PES participant households are older, have more formal education, have more land, and have more off-farm income than those households that do not participate (Zbinden and Lee, 2005; Arriagada et al., 2009; Jones et al., 2017). A growing body of PES research shows that the motivations to participate in these programs are complex and include a combination of

Corresponding author at: 238 Forestry Building, Colorado State University, Ft. Collins, CO 80523-1033, USA. E-mail address: [email protected] (K.W. Jones).

https://doi.org/10.1016/j.ecoser.2018.11.004 Received 25 July 2018; Received in revised form 9 November 2018; Accepted 9 November 2018 Available online 21 November 2018 2212-0416/ © 2018 Elsevier B.V. All rights reserved.

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

financial and non-financial reasons (Zanella et al., 2014; Mendez-Lopez et al., 2015; Grillos, 2017; Lansing, 2017; Moros et al., 2017). Financial motivators are in play when the in-kind or cash payment is the primary reason for participating in the program. Non-financial motivators include both pro-social and pro-environmental motivations (Deng et al., 2016; Grillos, 2017; Moros et al., 2017). Pro-social motivations arise from the need for social approval or reputation and occur when exposed to social norms that encourage a particular behavior (Granovetter, 1985). Pro-social motivations can also be related to the value one derives from contributing to the well-being of others, or altruism (Becker, 1974). Pro-environmental motivations can arise from intrinsic conservation values or from instrumental and relational value that people receive from nature (Pascual et al., 2017). Both intrinsic and instrumental pro-environmental motivations have been linked to participation in PES programs (Bottazzi et al., 2018). That non-financial motivators play a large role in determining who participates in PES programs may help explain why counterfactual impact evaluations of PES programs have found modest impacts, or no impacts, on material well-being indicators (Uchida et al., 2007; Hedge and Bull, 2011; Samii et al., 2014; Alix-Garcia et al., 2015; Arriagada et al., 2015; Börner et al., 2017; Blundo-Canto et al., 2018; Liu and Kontoleon, 2018). Previous literature attributes PES programs’ modest impact on material well-being to both small payment size and disproportionate participation in PES programs by households with more favorable socioeconomic profiles (Börner et al., 2017; Liu and Kontoleon, 2018). However, a third reason could be that households sign up for PES programs for pro-social or pro-environmental benefits in addition to or instead of for material benefits. There is increasing recognition that conservation interventions influence non-material outcomes and components of subjective well-being (Woodhouse et al., 2015), and these non-material benefits may drive an individual’s decision to participate in conservation programs. Only a handful of impact evaluations consider the social or non-material impacts of PES programs (Blundo-Canto et al., 2018). One exception to the latter is Arriagada et al. (2015) that measured changes in self-reported quality of life for Costa Rican households participating in PES programs and compared those changes to households not participating in PES programs; they found no statistical difference between the two. The goal of this study is to assess the linkages between why households participate in PHS programs and how that decision is related (or not) to the types of social and economic outcomes of these programs in Veracruz State, Mexico. First, we examine how the ability to participate in PHS programs and motivations to participate in PHS programs influence household enrollment. Second, we evaluate the impact of participation in PHS programs on material and non-material outcomes. We explore these two research objectives jointly because the financial and non-financial reasons households choose to enroll in PHS programs should be related to the types of benefits—material versus non-material—we observe for those households participating in the program. To date, the PES literature has paid little attention to the connections between what motivates a person to participate in PES programs and the type of environmental or social outcomes that the program achieves (a recent exception to this is Bottazzi et al., 2018). This connection is important to test since it has implications for the type of outcomes that should be expected and are measured in future PES program evaluations. In addition, our study examines whether and how land tenure influences participation in PHS programs and benefits received from participating in PHS programs. Mexico has a communal property rights system known as ejidos. Ejidal systems include two types of property rights: individual parcels managed by the household and shared-access common-use lands managed collectively by the ejido members (Assies, 2008; Alix-Garcia et al., 2015). Ejido members, known as ejidatarios, participate in assemblies and have voting rights. Within some ejidos, non-members or avecindados, have acquired individual parcels but often lack formal voting rights in the assemblies. PHS program contracts in

Mexico are made in at least three ways: (1) with ejido leaders for conservation on common-use lands; (2) directly with ejidatarios or avecindados to enroll their individual parcels in the PHS program; or (3) directly with private landowners to enroll their land in the PHS program. A handful of studies have examined how communities make decisions for group-level participation in collective PES program contracts (i.e., option one above) (e.g., Bremer et al., 2014; Murtinho and Hayes, 2017). Our study assesses household decision-making to enroll individual lands in PHS programs when that household is part of a communal land tenure system (i.e., option two above). We also focus on private household decision-making (i.e., option three above). This allows us to assess whether ejidatarios’ or avecindados’ decision to enroll their individual parcels in PHS programs and the benefits received are more akin to option one or option three above. Understanding the factors that influence participation in PHS programs and the types of benefits a household receives from PHS programs, across different land tenure systems, is important for the design and targeting of future PES programs, and ultimately the ability of these programs to meet their desired environmental and social objectives. 2. Methods 2.1. Study area Our study focuses on two PHS programs in central Veracruz State, Mexico. These PHS programs are organized around sub-watersheds (drainage basins), with downstream water consumers paying upstream land managers to maintain forest cover. Both PHS programs, and thus sub-watersheds, are located within the Antigua River Watershed (∼15,000 ha) (Fig. 1). These sub-watersheds overlap multiple municipalities. The study area is predominantly tropical montane cloud forest with pine-oak forest dominating at the highest elevations. About 65% of the original forest cover has been converted to agricultural landscapes, including cattle pasture, shade grown and intensive coffee, and subsistence crops such as maize and beans (Asbjornsen et al., 2017). More recently, cash crops such as sugar cane and potatoes are being planted within the watershed. These land cover changes directly influence downstream drinking water supply for the urban areas of Xalapa (∼450,000 persons) and Coatepec (∼80,000 persons). The first PHS program in Mexico started in the Gavilanes sub-watershed in 2002–2003 (Nava-López et al., 2018). This PHS program is administered by FIDECOAGUA (by its acronym in Spanish: Fidecomiso Coatepecano para la Conservación del Bosque y el Agua), a quasi-governmental agency dedicated to the management of the PHS program. A second PHS program started in 2008 targeting the Pixquiac sub-watershed. Referred to as PROSAPIX (by its acronym in Spanish: Programa de Compensación para Servicios Ambientales en la Cuenca del Río Pixquiac), this PHS program is administered by SENDAS (by its acronym in Spanish: Senderos y Encuentros para un Desarollo Autónomo Sustentable), a sustainable development non-profit that offers a variety of technical assistance and support for rural communities in the area (Nava-López et al., 2018). Both PHS programs receive funds from the Mexican national forestry commission, CONAFOR (by its acronym in Spanish: Comisión Nacional Forestal), and those funds are matched with monies raised locally, mostly through a surcharge on water bills in Coatepec and Xalapa. While the institutional structure of these PHS programs varies (Nava-López et al., 2018), both programs target forest conservation, pay landowners $1100 MXN pesos/ha (∼$60 USD/ha in 2015), and require 5-year contracts. About 2500 ha of forest in the study area (Fig. 1) was under PHS program contract as of 2015 (Nava-López et al., 2018). This represents contracts with approximately 90 private landowners and 120 ejidatarios or avecindados, as well as four collective PHS program contracts with ejidos for common-use forest. 33

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

Fig. 1. Location of the Pixquiac and Gavilanes sub-watersheds in Veracruz State, Mexico.

2.2. Data collection

program, and within each program, by upper and lower ecological zones within that sub-watershed. The latter was done because the upper sub-watershed tends to be pasture and the lower sub-watershed tends to be coffee production, affecting the opportunity costs of conservation. PHS program operators helped identify households participating in the PHS program and enumerators walked around the same communities to identify households not participating in the program. Ejidal households that were enrolled as part of an ejido common-use lands contract—that is, ejidal households where the common-use forestlands were enrolled in the PHS program—were not included in the sample used in this paper. The final survey sample used in this paper includes 186 households: 141 ejidal households (82 ejidal households in the PHS program and 59 ejidal households not in the PHS program) and 45 private households (25 private households in the PHS program and 20 private households not in the PHS program). In total, we surveyed approximately 50% of households enrolled in the PHS program in our study area (∼30% of private PHS enrollees and ∼70% of ejidal households that had enrolled their individual parcels). We surveyed one person from each household, focusing on key decision makers who were aware of land management decisions. The survey took an average of 50 minutes to complete. Survey data were entered into Excel1 and analyzed using the statistical software Stata.

We designed a mixed methods study that included 56 semi-structured interviews in 2015 and 186 household surveys in 2016 with individual ejidal (ejidatarios and avecindados) and private households in the study area. Both instruments collected information on: (1) demographic and socioeconomic characteristics of the households; (2) the decision to participate or not in the PHS programs; and (3) the benefits and outcomes resulting from the PHS programs. Both instruments were pre-tested in the field and final data collection was conducted in Spanish by native or advanced Spanish-speakers. To collect interview data, we used purposive and snowball sampling methods to contact interviewees (Bernard, 2011). Starting with key informants, such as non-governmental and community representatives, we interviewed people who managed land within both PHS programs and asked if interviewees could recommend neighbors, friends, relatives, and acquaintances who fit a similar description as them. Some land managers lived downstream, in a city. We conducted a total of 56 household interviews: 36 PHS program participants (about 17% of total participants) and 20 non-participants. Private individuals in PHS contracts and ejidal households that were in either an individual contract or an ejido common-use lands contract were included in the interviews. We only interviewed one person from each household. Thirty-one of the interviewees were from the Pixquiac sub-watershed while 25 were from the Gavilanes sub-watershed. Average time to complete the interview was 45 minutes. Interviews were transcribed into Spanish and were thematically coded using NVivo qualitative data management software. To collect survey data, we first stratified our study area by PHS

1 Access to the survey data used in analysis can be found on figshare (https:// figshare.com/articles/Veracruz_Mexico_Payment_for_Ecosystem_Services/ 7294100) or by request from the authors.

34

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

2.3. Data analysis In the interviews and surveys, we asked respondents to state their primary reasons for participating in the PHS programs. Responses were unprompted and later coded into pro-environmental, pro-social, and financial motivations. In the interviews, we asked a number of additional descriptive questions to participants about the types of benefits they receive from being enrolled in the PHS programs. These questions get at motivations indirectly in that they capture the financial and nonfinancial reasons for participating in the forest conservation program and help characterize whether the financial or non-financial aspects and benefits of the program drive participation. Using the household survey data, logistic regression was used to examine correlations between household characteristics and participation in the PHS programs. The counterfactual impact evaluation method of matching was used to measure the effect of participation on material and non-material benefits. These statistical analyses are described in more detail below.

To capture social capital and to proxy for pro-social motivations we measured the existence of community sanctions (1/0) as an indicator of strong community organization and social norms. Many communities have pre-existing rules and norms regarding resource management in our study area; thus, the presence of community sanctions was expected to positively correlate with PHS program participation, especially among ejidal households. As an additional measure of social motivations, we measured whether the household participated in community or ejidal organizations; recorded as “1” for yes and “0” for no. This captures an action or behavior of the household and serves as a proxy for pro-social motivations. To capture pro-environmental motivations, we collected information on household participation in an environmental group or organization. Participation was recorded as “1” for yes and “0” for no. This variable captures a behavior of the household and is used as an indicator for pro-environmental motivations; however, it does not allow differentiation between instrumental and intrinsic environmental motivations.

2.3.1. Analyzing participation in PHS programs To explore how factors that might affect ability to participate, and proxies for motivations to participate, were correlated with household participation in PHS programs we first used two-sample Wilcoxon rank-sum tests and twosample t-tests to calculate differences in means in independent variables across PHS program participants and non-participants (Vaske, 2008). We did this for all households and separately for ejidal and private households. Second, we used logistic regression to analyze the factors correlated with participation. Prior to regression analysis we tested for multicollinearity between independent variables using tolerance scores and variance inflation factors (Vaske, 2008). We compared goodness-of-fit across several specifications of regression models using McKelvey and Zavoina Pseudo R-square and the Hosmer-Lemeshow goodness-of-fit statistic (Hosmer et al., 2013). We estimated the final logistic model for the full sample of households and separately for ejidal and private households to test whether there were differences across these land tenure categories. While the total number of private households is small (N = 45), a sample size of > 40 is considered acceptable for most statistical analyses (Vaske, 2008). We report marginal effects and robust standard errors for all regressions. Our dependent variable in statistical analysis was participation in PHS programs and was measured in our household surveys as a “1” if the household had ever participated and “0” otherwise. We measured a number of independent variables in the survey instrument expected to influence a household’s ability to participate in a PHS program. Household characteristics including age and formal education were expected to positively influence participation since they indicate higher human capital. Education was converted to a “0” for no schooling and a “1” for primary schooling or higher, given the limited number of years of schooling in our study area. The households’ total hectares of land were recorded and was log-transformed to correct for skewness in the regression model; this variable was expected to correlate positively with participation. We also asked households if any land on their parcel was too steep to farm, recorded as a “1” for yes or “0” for no. This variable was expected to be positively correlated with participation since it would indicate less productive natural capital for farming or ranching, which are the two main livelihood activities in the area. However, steep land could still be productive for other uses such as tourism or non-extractive forest activities. To measure financial capital, we recorded the total wages of all household members; we asked about any cash income through farm or off-farm activities. Income was recorded in pesos and converted to 2015 US dollars (USD) and logtransformed to correct for skewness in the regression model. We measured physical capital through ownership of 11 durable goods: car, motorcycle; scooter; bicycle; cellular phone; television; electricity; flushing toilet; refrigerator; washing machine; and gas stove. We created an asset count summing all 11 goods, giving equal weight to each asset.

2.3.2. Measuring the benefits of PHS programs We used the quasi-experimental impact evaluation method of matching to test whether there were differences in material and nonmaterial benefits across households participating in the PHS programs and those not participating in the programs. Matching helps reduce observable selection bias between treatment and control observations (Rubin, 2006; Imbens and Wooldridge, 2009). Matching selects the sample of households that are most similar in observable characteristics, and then compares average outcomes across the treated and control groups. In our study, these observable characteristics come from the household survey. The identifying assumption in matching is that selection bias between those in the program and those not in the program can be controlled for with these observable characteristics and that observations are not subject to hidden, or unmeasured, bias (Imbens and Wooldridge, 2009). While less robust than quasi-experimental methods that use before and after data, matching provides estimates of treatment effects that are similar to these other methods if hidden bias is not a concern (Jones and Lewis, 2015). We measured two material well-being outcomes in our household survey: change in durable goods and change in cattle ownership. Since we lacked baseline data from our study area, we asked each respondent to recall these values in 2010 and in 2015. We chose 2010 as the recall year since there was a large gubernatorial election in that year and used this cue to reference the time frame we were asking about. However, there could still be measurement error due to the retrospective nature of the questions (Mullan et al., 2013). For durable goods we used the sum of the 11 assets described above and calculated the difference in ownership between 2015 and 2010. Cattle ownership was recorded as two separate binary variables for dairy and meat cattle. We summed these two variables to create a count variable ranging between zero and two and took the difference between whether either type of cattle was owned in 2015 and 2010. To measure non-material benefits from the PHS program we asked households to report changes in their (1) individual quality of life and (2) community quality of life. Specifically, respondents answered whether quality of life had improved (recorded as “1”), decreased (“−1”), or stayed the same (“0”) between 2015 and 2010. The selfreported subjective well-being question for individual households was taken from Arriagada et al.’s (2015) evaluation in Costa Rica; we added a self-reported measure for community quality given the importance of the ejidal land tenure system in our study area. We used nearest neighbor covariate matching, with bias-adjustment and robust standard errors to evaluate the impact of the PHS programs on these four outcome variables (Abadie et al., 2004; Abadie and Imbens, 2006). We present matching results using two different metrics—inverse sample standard errors and Mahalanobis metric—and with replacement. We matched participants to one, three, and six 35

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

Table 1 Summary statistics for household survey. Variable (Measured)

Table 2 Logistic regression of factors that influence PHS program participation.

All households Ejidal households

Age (continuous)

Private households

Variable

All households

Ejidal households

Private households

0.007*** (0.002) 0.067 (0.077) 0.082** (0.041) 0.062 (0.076) 0.003 (0.011) 0.043** (0.016) 0.224*** (0.072) −0.038 (0.082) 0.142* (0.082)

0.005* (0.003) 0.041 (0.089) 0.108** (0.054) 0.050 (0.092) −0.014 (0.016) 0.031* (0.021) 0.214** (0.088) −0.024 (0.100) 0.085 (0.102)

0.013** (0.004) 0.019 (0.127) 0.009 (0.077) 0.335** (0.144) 0.042** (0.017) 0.037 (0.026) 0.206 (0.172) −0.055 (0.200) 0.422** (0.180)

0.20 0.82

0.15 0.61

0.60 0.91

183

141

42

53.3 (14.6) Education (0/1) 0.53 (0.50) Hectares (continuous) 7.48 (7.72) Steep land (0/1) 0.67 (0.47) Annual total income (USD 2015) 1808 (1988) Durable goods count (0–11) 4.24 (2.13) Community sanctions (0/1) 0.59 (0.49) Community group (0/1) 0.46 (0.50) Environmental group (0/1) 0.33 (0.47)

53.6 (14.67) 0.56 (0.50) 6.86 (6.55) 0.67 (0.47) 1928 (2077) 4.25 (2.02) 0.68 (0.46) 0.49 (0.50) 0.36 (0.48)

52.3 (14.62) 0.44 (0.50) 9.44 (10.42) 0.69 (0.47) 1402 (1609) 4.18 (2.45) 0.31 (0.46) 0.37 (0.48) 0.24 (0.43)

Age

Observations

141

45

McKelvey and Zavonia’s R2 P-value from Hosmer and Lemeshow goodness-of-fit test1 Observations

186

Education Hectares (Log-transformed) Steep land Annual total income (Log-transformed) Durable goods count Community sanctions Community group Environmental group

Note: Mean value reported with standard deviation in parentheses.

households with the most similar characteristics, to make sure our results were not sensitive to the number of nearest neighbors. Since results were qualitatively similar, we only report matches with six nearest neighbors below. For each outcome variable we determined a best set of covariates to control for in the matching based on the factors correlated with participation in the PHS programs in the logistic regression analysis and variables that could influence the outcome of interest (the specific variables for each outcome are identified in Section 3 Results). We present the average treatment effect for the treated using the full sample of observations, and separately by ejidal and private households to test for differences in these land tenure categories.

Note: Marginal effects reported with robust standard errors in parentheses; ***pvalue < 0.01; **p-value < 0.05; *p-value < 0.10. 1 A good fit according to Hosmer and Lemeshow goodness-of-fit test is reflected in a large p-value (Hosmer et al., 2013).

in PHS programs when we separated ejidal and private households (Table 2). For ejidal households, age, number of hectares managed by the household, wealth and community sanctions were statistically significant with a positive effect on enrollment. For private households, age remained statistically significant; presence of steeply sloped land was related to enrollment but total hectares managed was not; total household income was positively correlated with PHS program participation but durable goods was insignificant; and environmental motivations had a positive and statistically significant influence on participation but social factors were not statistically significant.

3. Results 3.1. Participation in PHS programs In our household survey, respondents were, on average, 53 years old, with about 50% having at least a primary school education (Table 1). About 70% of our respondents were male. Households had an average land size of seven hectares. The annual mean cash income reported was around $1800 USD and was slightly higher for ejidal households than private households (but not statistically different). Out of 11 durable goods, households owned an average of four. About 60% of households stated that there were community sanctions. About 46% of our sample participated in a community group and 33% participated in an environmental group. Using differences in means tests, households enrolled in the PHS programs were on average older; owned more land; had more land with steep slopes; and were more likely to live in a community that had sanctions (Appendix A). There were no statistically significant differences between PHS program participant and non-participant households across income or durable goods. Participation in an environmental group was statistically different and higher for PHS program households. Although participation in a community organization was higher for PHS program households, the difference was not statistically significant. Using logistic regression and the full sample, age, number of hectares, the count of durable goods owned, community sanctions and participation in an environmental group were the statistically significant variables in explaining who participated in PHS programs (Table 2). Each variable had a positive influence on participation. We found differences in the factors influencing the decision to participate

3.2. Benefits of PHS programs Using differences in means tests to compare households enrolled in the PHS programs and households not enrolled in the PHS programs for changes in material and non-material benefit indicators, we found statistically significant differences for non-material benefits but not material benefits in the full sample of households (Appendix B). Specifically, PHS program participants were more likely to report that their household and community quality of life had improved in the last five years compared to non-PHS program participants. When separated out by ejidal and private households, these results were statistically significant only for private households (Appendix B). However, since some household characteristics were statistically different across PHS program participant and non-participant households (Appendix A), these naïve differences in means tests are subject to bias due to the type of household that self-selected into the PHS programs. Matching provides a more robust comparison by only selecting those households with similar covariate balance. The improved covariate balance following matching is seen in Appendix C. Some differences remain and the bias-adjustment in the nearest neighbor matching program helps control for these remaining differences (Abadie et al., 2004). Using nearest neighbor matching, we found no statistically significant differences in changes in durable goods or changes in cattle ownership between the full sample of households participating in the PHS programs and non-participants (Table 3). The observable 36

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

Table 3 Effect of PHS program participation on material and non-material benefits. Outcome

Change in durable goods1 Mahalanobis metric Inverse sample standard errors metric Change in cattle1 Mahalanobis metric Inverse sample standard errors metric Change in household quality of life2 Mahalanobis metric Inverse sample standard errors metric Change in community quality2 Mahalanobis metric Inverse sample standard errors metric Observations

PHS versus nonPHS program participants

Ejidal households: PHS versus non-PHS program participants

Private households: PHS versus non-PHS program participants

0.05 (0.17) 0.03 (0.17)

0.26 (0.20) 0.26 (0.20)

−0.21 (0.39) −0.13 (0.41)

0.09 (0.08) 0.10 (0.07)

0.03 (0.08) 0.01 (0.08)

0.47*** (0.14) 0.47*** (0.15)

0.26** (0.12) 0.26** (0.12)

0.16 (0.13) 0.13 (0.14)

0.61** (0.22) 0.54** (0.23)

0.33** (0.13) 0.30** (0.13)

0.32** (0.15) 0.30** (0.15)

0.16 (0.23) 0.24 (0.24)

185

141

45

Note: Nearest neighbor covariate matching with replacement and bias-adjustment was used to estimate average treatment effect for the treated. Coefficient is presented with robust standard errors in parentheses. Matching with six nearest neighbors reported; results were qualitatively similar for one and three nearest neighbors. 1 Covarites included age, household total income in 2010, and total hectares of land managed in 2010, and the 2010 value of the variable (e.g., 2010 durable goods count for change in durable goods). 2 Covarites included age, household total income in 2010, total hectares of land managed in 2010, the 2010 count of durable goods owned, and presence of community sanctions.

participating in the PHS programs during interviews most respondents mentioned “to protect the forests” (68%) while slightly less stated “the financial payment” (61%). About 35% of those interviewed mentioned both forest conservation and financial payment. Only a handful (10%) brought up the influence of social connections, such as ejidal leaders, neighbors, or family members, in the open-ended interviews. However, in the household survey, 33% of participants stated forest conservation, 36% mentioned the financial incentives, and 23% noted the influence of a family member, neighbor or local ejido leader, as their primary motivation for participation. These percentages were similar across private and ejidal households. The alignment of the PHS program with existing pro-environmental motivations was expressed in many of our interviews. One respondent highlighted how the program enabled their intrinsic values toward forests as: “Yes, the truth is that I’m interested in the program because, as I say, well, natural [things] that have existed many years interest me. And we [risk] losing it, right?” (Interviewee 72). Whereas others expressed instrumental reasons for entering the program as: “[the program] helps us know how to care for the forest…and that allows us to help ourselves” (Interviewee 11). The environmental benefits of participating in the PHS programs were also readily expressed by many people in the interviews. For example, one person stated: “There’s that spring that doesn’t dry up, right? So we have more water” (Interviewee 44). One forward-thinking interviewee responded that they benefited from the PHS program because it had helped secure environmental benefits for future generations. They said:

characteristics used in matching included: age, household total income in 2010, total hectares of land managed in 2010, and the baseline value (2010 value) of the outcome variable. We did find that private households enrolled in the PHS programs were more likely to increase cattle ownership when compared to private households not enrolled in the PHS programs. Specifically, a private household enrolled in the PHS programs increased cattle ownership by an average of 0.5%-points more than households not in the PHS programs between 2015 and 2010 (p-value < 0.01). These results were robust to including additional covariates in the matching estimator and changing the number of nearest neighbors used in the match. We found statistically significant differences between PHS program participants and non-participants in self-reported changes in household quality of life and community quality between 2015 and 2010 (pvalue < 0.05; Table 3). The observable characteristics used in the matching included: age, household total income in 2010, total hectares of land managed in 2010, the 2010 count of durable goods owned, and the presence of community sanctions. PHS program participants were more likely to state that both household and community quality of life had improved in the last five years compared to households not participating in the PHS programs. The treatment effect for both indicators was around 0.3%-points. When separated by household type, we found that private households participating in PHS programs were more likely to perceive a positive change in their household well-being while ejidal households participating in the PHS programs were more likely to perceive a positive change in their community quality. These results were robust to including additional covariates in the matching estimator and using different numbers of control observations in the nearest neighbor match.

Suppose that I planted [trees] 5 years ago, planted, let's say, 100 trees. I am going to take care of them because I… [eventually] I am going to die later. …And the trees that you cut down, [if] you did not plant them, you are taking away from other people who aren’t yet alive. Now you have to plant trees to provide for those who come [after you]. (Interviewee 63)

3.3. Stated reasons for participating in and benefits from PHS programs When individuals were asked about their primary reasons for 37

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

Interview respondents also expressed financial motivations for participating in the PHS programs. For example, the financial security of participating in the program was expressed as “Although [the payment] is small, at least they give it to us” (Interviewee 77), and a reduced need to travel or migrate to find employment:

intrinsic values toward nature were expressed in our interviews. In logistic regression analysis, the effect of environmental behaviors on PHS program participation was most pronounced in our sub-set of private households. Pro-social motivations, operationalized through the presence of community sanctions, had a positive influence on ejidal households participating in the PHS programs on their individual parcels. Other studies have found that: community organization positively explains participation of communities in collective PES program contracts (Bremer et al., 2014; Murtinho and Hayes, 2017; Jones et al., 2018); pro-social motivations encourage PES program participation of private households (Grillos, 2017); and participation in PES programs is linked to social pressure from neighbors (Deng et al., 2016). Thus, in our study, ejidal households enrolling their individual parcels in the PHS program behave differently than private households in terms of the non-financial motivators influencing participation. The motivators affecting ejidal households were more similar to those that influence communal decisions to enroll in collective PES programs (e.g., Bremer et al., 2014; Murtinho and Hayes, 2017; Jones et al., 2018), rather than the pro-environmental motivations that were related to private household enrollment in this study.

Before, we lived off the land, taking [trees] from the town of Coatepec. We lived as workers…traveling [wherever] the wood from Coatepec [went]. But with this program, we’ve had, well, we don’t do it anymore. We’ve improved our forest monitoring and we don’t have to work like we did before. (Interviewee 10) Respondents also expressed that participating in the PHS programs provided a more secure livelihood strategy for older households: “[my father] is now elderly and he barely works, and so they give him a little [support] and, well, he goes on taking care of himself” (Interviewee 47). Many interview respondents expressed that the PHS program payment was too low, even PHS participants, despite their voluntary participation. Most PHS program participants also reported that they used the cash from the PHS programs primarily to meet immediate consumption needs. This was exemplified by an interviewee: “We [use the money for] what’s necessary for the house: all of the family’s clothing, if there’s illness or whatever’s needed. There’s always something” (Interviewee 118). Other uses of the money that were mentioned, albeit much less frequently, included maintaining their forest and for agricultural production.

4.2. Benefits of PHS programs Our results also parallel the growing number of impact evaluations that find that the material benefits from PES programs are modest at best (Samii et al., 2014; Alix-Garcia et al., 2015; Arriagada et al., 2015; Börner et al., 2017; Liu and Kontoleon, 2018). The modest impacts of the PHS programs in our study on material well-being (Table 3) are likely due to both the small payment size and the non-financial reasons that motivated households to participate in the programs. Average reported payment received by PHS program participants in 2015 was $170 USD total per year. Most households stated that this payment was too low and that they primarily used this money to support daily expenditures. Given these statements, it is not surprising that for the average household we observed few statistically significant changes in material well-being indicators. We did find a statistically significant and positive change in cattle production for private households enrolled in PHS programs; a similar effect was found in impact evaluations of PES programs in Costa Rica (Arriagada et al., 2015) and Mexico (Alix-Garcia et al., 2015). While cattle, and livestock more generally, are a form of wealth for rural households, this impact could also lead to environmental leakages, undermining additional ecosystem services benefits created by PES programs. While Arriagada et al.’s (2015) impact evaluation of Costa Rica’s PES program finds no statistically significant change in stated quality of life for participants relative to non-participants, the authors conclude that PES programs likely confer non-material benefits to participants. Our results support this assessment of non-material benefits from participating in financial incentive programs in both qualitative interviews and our impact evaluation estimates (Table 3). Private households enrolled in the PHS programs were more likely to state that their household quality of life improved relative to households not participating in PHS programs. The intrinsic, instrumental, or relational benefits received from forest conservation could be one explanation for why this indicator for non-material benefits may have increased relative to nonparticipants. Ejidal households enrolled in the PHS programs, however, were more likely to perceive an increase in the quality of their community over the last five years versus non-participants. Participating in the PHS programs may have strengthened their community social relations and trust (e.g., Alix-Garcia et al., 2018), which translated to a perceived increase in the quality of life within their community. In general, both types of households mentioned non-material benefits as important reasons for their participation.

4. Discussion 4.1. Participation in PHS programs A growing body of research finds that participation in PES programs is explained by both the ability to participate and a combination of financial and non-financial motivators (Mendez-Lopez et al., 2015; Grillos, 2017; Lansing, 2017; Moros et al., 2017; Raes et al., 2017). This study supports these findings and contributes important insights about the differences in determinants of participation in PHS programs across private and ejidal households in Mexico. Similar to other studies on PES program participation (e.g., Zbinden and Lee, 2005; Arriagada et al., 2009; Lansing, 2017), we find that human and natural capital affect the ability to participate in PHS programs. Specifically, we find that age influences participation with older households more likely to sign up for conservation programs. Older households may find forest conservation more attractive given the physical demands of farming (Bremer et al., 2014; Jones et al., 2017). Natural capital affects the opportunity costs of conservation. We find that having more land is positively correlated with the decision to enroll in PHS programs and that having land that was less suitable for farming or ranching (steep slopes) leads to enrollment in the PHS programs. Financial and physical capital – measures of household wealth – were also important determinants of participation in our PHS programs. Similar to other PES studies (e.g., Zbinden and Lee, 2005; Arriagada et al., 2009; Lansing, 2017), we find that households with more wealth, measured either through durable goods or total income, were more likely to participate in the PHS programs. Wealthier households may face fewer financial barriers to enrollment and have off-farm opportunities that allow them to set aside land for conservation. Non-financial motivators also had a large effect on determining enrollment in the PHS programs in our study area. Previous studies find a positive relationship between environmental values and participation in PES programs (Kosoy et al., 2008; Scullion et al., 2011; Bremer et al., 2014; Zanella et al., 2014; Deng et al., 2016; Figueroa et al., 2016; Jones et al., 2017; Raes et al., 2017). We find that pro-environmental motivations are the most stated reason for participating in the PHS programs in our qualitative interviews. A mix of instrumental and 38

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

4.3. Policy implications

reinforcing environmental values, reducing financial risk and uncertainty, and strengthening social networks, exist for PHS program participants. Research on other PES programs document similar nonmaterial benefits and also highlight the role PES programs can play in increasing tenure security (e.g., Arriagada et al., 2009; Hejnowicz et al., 2014; Jones et al., 2017). Our results support the notion that non-material benefits likely influence subjective components of well-being and are thus an important dimension of human well-being to assess in conservation impact evaluations (Woodhouse et al., 2015). Interestingly, the type of non-material benefits realized also varied by land tenure category. Ejidal households participating in the PHS programs were more likely to report positive changes in community quality of life versus household-level quality of life. Thus, this dimension of community or social capital is important to consider within communal land tenure systems, even when participation decisions are made at the household level. Participatory methods that allow local people to define quality of life domains and relate these domains directly to conservation interventions could be used in future PES program evaluations to complement top-down metrics like the ones used in this study (e.g., Gross-Camp, 2017; Rasolofoson et al., 2018).

There is a direct link between the type of individual (or community) that participates in PES programs and the environmental or social outcomes that can be achieved. This has important implications for the future design and evaluation of PES programs. In particular, the current logic about PES programs suggests that increasing the size of payment, or better targeting of payments to high deforestation risk areas, will increase additionality of PES programs (Wunder, 2006; Engel et al., 2008; Wunder, 2015; Börner et al., 2017). This logic holds if financial incentives motivate participation; however, if non-financial motivators are an important factor, then a linear relationship between payment level and enrollment will not exist. Other barriers to participation have also been identified in the PES literature, including lack of access to alternative livelihoods, credit constraints, lack of trust in the implementing agency, and insecure land tenure (Kosoy et al., 2008; Jayachandran, 2013; Bremer et al., 2014; Jones et al., 2017). More attention to the complexity of factors that influence behavior change can help inform future PES program designs to take advantage of both financial and non-financial motivators. In some cases, using pro-social or pro-environmental motivations as a complement to financial motivators may enhance conservation outcomes (e.g., Grillos, 2017; Moros et al., 2017). In our study, we find that the type of non-financial motivator that is important varies by land tenure category and thus the design of PES programs may also need to vary across these two systems. Additionally, there is a need to better document the non-material benefits that PES programs provide and articulate the causal relationships that exist between PES programs and these benefits in impact evaluation. Our study suggests that non-material benefits in the form of

Acknowledgements The authors would like to acknowledge NSF’s Dynamics of Coupled Natural-Human Systems (CNH) program grant no. 1313804 for funding this research. This research would not have been possible without the collaboration and cooperation of staff at FIDECOAGUA and SENDAS, and the households that graciously participated in this study.

Appendix A See Table 4

Appendix A Differences in means for household characteristics across PHS program participants and non-participants. Variable

Age Education Hectares Steep land Durable goods count Annual total income (USD 2015) Community sanctions Community group Environmental group Observations

All households

Ejidal households

Private households

PHS program participant

Not in PHS program

PHS program participant

Not in PHS program

PHS program participant

Not in PHS program

55** (14) 0.57 (0.50) 8.35* (7.57) 0.7* (0.4) 4.4 (2.1) 1749 (1861) 0.7** (0.5) 0.5 (0.5) 0.4*** (0.5)

51 (14) 0.49 (0.50) 6.42 (7.80) 0.6 (0.5) 4.1 (2.1) 1877 (2136) 0.5 (0.5) 0.4 (0.5) 0.2 (0.4)

56* (15) 0.57 (0.50) 7.43* (6.13) 0.7 (0.5) 4.4 (2.0) 1833 (1933) 0.7* (0.4) 0.5 (0.5) 0.4* (0.5)

51 (15) 0.54 (0.50) 5.61 (5.39) 0.6 (0.5) 4.3 (2.0) 2059 (2272) 0.6 (0.5) 0.5 (0.5) 0.3 (0.5)

55 (14) 0.55 (0.50) 10.88 (10.27) 0.8* (0.4) 4.3 (2.5) 1320 (1411) 0.5 (0.5) 0.5 (0.5) 0.5** (0.5)

50 (14) 0.36 (0.50) 8.05 (11.11) 0.6 (0.5) 3.7 (2.2) 1454 (1750) 0.3 (0.5) 0.3 (0.5) 0.1 (0.3)

102

84

82

59

20

25

Note: Mean value reported with standard deviation in parentheses. Differences in means calculated using two-sample Wilcoxon rank-sum test for dichotomous variables and two-sample t-test with unequal variances for continuous variables. Statistical difference reported as: ***p-value < 0.01; **p-value < 0.05; *pvalue < 0.10.

39

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

Appendix B See Table 5 Appendix B Differences in means for material and non-material benefits. Variable

All households

Change in durable goods Change in cattle Change in household quality of life Change in community quality of life

Ejidal households

Private households

PHS program participant

Not in PHS program

PHS program participant

Not in PHS program

PHS program participant

Not in PHS program

1.12 (0.14) 0.09 (0.05) 0.50* (0.07) 0.46** (0.07)

1.17 (0.14) 0.07 (0.05) 0.29 (0.09) 0.20 (0.09)

1.04 (0.16) 0.01 (0.05) 0.50 (0.08) 0.46 (0.09)

0.87 (0.13) 0.10 (0.07) 0.44 (0.10) 0.24 (0.11)

1.33 (0.29) 0.29** (0.11) 0.50** (0.11) 0.46* (0.11)

1.72 (0.27) 0.00 (0.07) 0.00 (0.16) 0.11 (0.15)

Note: Mean value reported with standard deviation in parentheses. Differences in means calculated using two-sample t-test with unequal variances for all variables. Statistical difference reported as: ***p-value < 0.01; **p-value < 0.05; *p-value < 0.10.

Appendix C See Tables 6–9 Table C1 Covariate balance change in durable goods. Variable

All households

Age Hectares Baseline durable goods count Annual total income

Ejidal households

Private households

Before matching

After matching

Before matching

After matching

Before matching

After matching

0.34 0.33 0.15 0.06

0.13 0.13 0.06 −0.01

0.33 0.39 0.06 −0.13

0.15 0.20 0.01 −0.11

0.35 0.23 0.40 0.40

0.12 0.19 0.23 0.03

Note: Values represent covariate balance; a perfectly balanced covariate has a standardized difference of zero. Test implemented using command “tebalance summarize” in Stata 14. Results shown for 6 nearest neighbor matching and inverse sample standard errors metric.

Table C2 Covariate balance change in cattle. Variable

Age Hectares Annual total income Baseline cattle

All households

Ejidal households

Private households

Before matching

After matching

Before matching

After matching

Before matching

After matching

0.34 0.04 0.33 0.07

0.13 −0.01 0.18 0.13

0.33 −0.13 0.39 0.01

0.16 −0.10 0.25 0.11

0.36 0.30 0.25 0.09

0.12 0.20 0.23 0.23

Note: Values represent covariate balance; a perfectly balanced covariate has a standardized difference of zero. Test implemented using command “tebalance summarize” in Stata 14. Results shown for 6 nearest neighbor matching and inverse sample standard errors metric. Table C3 Covariate balance change in household quality of life. Variable

Age Hectares Annual total income Baseline durable goods count Community sanctions

All households

Ejidal households

Private households

Before matching

After matching

Before matching

After matching

Before matching

After matching

0.34 0.34 0.04 0.14 0.38

0.18 0.21 −0.06 0.10 0.02

0.31 0.42 −0.12 0.03 0.30

0.24 0.20 −0.17 0.07 −0.02

0.40 0.22 0.35 0.42 0.36

0.18 0.22 0.09 0.31 0.35

Note: Values represent covariate balance; a perfectly balanced covariate has a standardized difference of zero. Test implemented using command “tebalance summarize” in Stata 14. Results shown for 6 nearest neighbor matching and inverse sample standard errors metric.

40

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al.

Table C4 Covariate balance change in community quality. Variable

Age Hectares Annual total income Baseline durable goods count Community sanctions

All households

Ejidal households

Private households

Before matching

After matching

Before matching

After matching

Before matching

After matching

0.34 0.34 0.04 0.14 0.38

0.18 0.21 −0.06 0.10 0.02

0.31 0.42 −0.12 0.03 0.30

0.24 0.20 −0.17 0.07 −0.02

0.40 0.22 0.35 0.42 0.36

0.14 0.28 0.04 0.08 0.35

Note: Values represent covariate balance; a perfectly balanced covariate has a standardized difference of zero. Test implemented using command “tebalance summarize” in Stata 14. Results shown for 6 nearest neighbor matching and inverse sample standard errors metric.

Appendix D. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.ecoser.2018.11.004.

17, 24–32. Gross-Camp, N., 2017. Tanzania’s community forests: their impact on human wellbeing and persistence in spite of the lack of benefit. Ecol. Soc. 22 (1), 37. Hedge, R., Bull, G.Q., 2011. Performance of an agro-forestry based payments-for-environmental-services project in Mozambique: a household level analysis. Ecol. Econ. 71, 122–130. Hejnowicz, A.P., Raffaelli, D.G., Rudd, M.A., White, P.C.L., 2014. Evaluating the outcomes of payments for ecosystem services programmes using a capital asset framework. Ecosyst. Serv. 9, 83–97. Hosmer, D.W., Lemshow, S., Sturdivant, R.X., 2013. Applied Logistic Regression, third ed. Wiley Publishers. Imbens, G.M., Wooldridge, J.M., 2009. Recent developments in the econometrics of program evaluation. J. Econ. Lit. 47, 5–86. Jayachandran, S., 2013. Liquidity constraints and deforestation: the limitations of payments for ecosystem services. Am. Econ. Rev. 103 (3), 309–313. Jones, K.W., Lewis, D.J., 2015. Estimating the counterfactual impact of conservation programs on land cover outcomes: the role of matching and panel regression techniques. PLoS One 10 (10). https://doi.org/10.1371/journal.poine.0141380. e0141380. Jones, K.W., Holland, M.B., Naughton-Treves, L., Morales, M., Suarez, L., Kennan, K., 2017. Forest conservation incentives and deforestation in the Ecuadorian Amazon. Environ. Conserv. 44 (1), 56–65. Jones, K.W., Brenes, C.M., Shinbrot, X., Lopez, W., Rivera-Castaneda, A., 2018. The influence of cash and technical assistance on household-level outcomes in payment for hydrological services programs in Chiapas, Mexico. Ecosyst. Serv. 31, 208–218. Kosoy, N., Corbera, E., Brown, K., 2008. Participation in payments for ecosystem services: case studies from the Lacandon rainforest, Mexico. Geoforum 39, 2073–2083. Lansing, D.M., 2017. Understanding smallholder participation in payments for ecosystem services: the case of Costa Rica. Human Ecol. 45, 77–87. Liu, Z., Kontoleon, A., 2018. Meta-analysis of livelihood impacts of payments for environmental services programmes in developing countries. Ecol. Econ. 149, 48–61. Martin-Ortega, J., Ojea, E., Roux, C., 2013. Payments for water ecosystem services in Latin America: a literature review and conceptual model. Ecosyst. Serv. 6, 122–132. Mendez-Lopez, M.E., Garcia-Frapolli, E., Ruiz-Mallen, I., Porter-Bolland, L., Reyes-Garcia, V., 2015. From paper to forest: local motives for participation in different conservation initiatives. Case studies in Southeastern Mexico. Environ. Manage. https:// doi.org/10.1007/s00267-015-0522-0. Moros, L., Velez, M.A., Corbera, E., 2017. Payments for ecosystem services and motivational crowding in Colombia’s Amazon Peidmont. Ecol. Econ. https://doi.org/10. 1016/j.ecolecon.2017.11.032. Mullan, K., Sills, E., Bauch, S., 2013. The reliability of retrospective data on asset ownership as a measure of past household wealth. Field Methods. https://doi.org/10. 1177/1525822X13510370. Murtinho, F., Hayes, T., 2017. Communal participation in payment for environmental services: Unpacking the collective decision to enroll. Environ. Manage. 59, 939–955. Nava-López, M., Selfa, T., Cordoba, D., Pischke, E.C., Torrez, D., Ávila-Foucat, S., Halvorsen, K., Maganda, C., 2018. Decentralized payments for watershed services programs in Veracruz, Mexico: challenges and implications for conservation. Nat. Resour. Soci. https://doi.org/10.1080/08941920.2018.1463420. Pagiola, S., Arcenas, A., Platais, G., 2005. Can payments for environmental services help reduce poverty? an exploration of the issues and the evidence to date from Latin America. World Dev. 33 (2), 237–253. Pascual, U., Balvanera, P., Diaz, S., Pataki, G., Roth, E., Stenseke, M., Watson, R.T., Dessane, E.B., Islar, M., Kelemen, E., Maris, V., Quaas, M., Subramanian, S.M., Wittmer, H., Adland, A., Ahn, S., Al-Hafedh, Y.S., Amankwah, E., Yagi, N., 2017. Valuing nature’s contributions to people: the IPBES approach. Curr. Opin. Environ. Sustainability 26, 7–16. Raes, L., Speelman, S., Aguirre, N., 2017. Farmers’ preferences of PES contracts to adopt silvopastoral systems in southern Ecuador, revealed through a choice experiment. Environ. Manage. https://doi.org/10.1007/s00267-017-0876-6. Rasolofoson, R.A., Nielsen, M.R., Jones, J.P.G., 2018. The potential of the Global Person Generated Index for evaluating the perceived impacts of conservation interventions

References Abadie, A., Durkker, D., Herr, J.L., Imbens, G.W., 2004. Implementing matching estimators for average treatment effects in Stata. Stata J. 4 (3), 290–311. Abadie, A., Imbens, G.W., 2006. Large sample properties of matching estimators for average treatment effects. Econometrica 74 (1), 235–267. Alix-Garcia, J.M., Sims, K.R., Yanez-Pagans, P., 2015. Only one tree from each seed? Environmental effectiveness and poverty alleviation in Mexico’s Payments for Ecosystem Services Program. Am. Econ. J.: Econ. Policy 7 (4), 1–40. Alix-Garcia, J.M., Sims, K.R.E., Olvera, V.H.O., Costica, L., Medina, J.D.F., Monroy, S.R., 2018. Payments for environmental services supported social capital while increasing land management. Proc. Natl. Acad. Sci. 115 (27), 7016–7021. Arriagada, R.A., Sills, E.O., Pattanayak, S.K., Ferraro, P.J., 2009. Combining qualitative and quantitative methods to evaluate participation in Costa Rica’s program of payments for environmental services. J. Sustainable For. 28, 343–367. Arriagada, R.A., Sills, E.O., Ferraro, P.J., Pattanayak, S.K., 2015. Do payments pay off? Evidence from participation in Costa Rica’s PES program. PLoS One 10 (7), 1–17. Asbjornsen, H., Manson, R.H., Scullion, J.J., Holwerda, F., Munoz-Villers, L.E., AlvaradoBarrientos, M.S., Geissert, D., Dawson, T.E., McDonnell, J.J., Bruijnzeel, L.A., 2017. Interactions between payments for hydrolgocial services, landowner decisions, and ecohydrological consequences: synergies and disconnection in the cloud forest zone of central Veracruz, Mexico. Ecol. Soc. 22 (2), 25. Assies, W., 2008. Land tenure and tenure regimes in Mexico: an overview. J. Agrarian Change 8 (1), 33–63. Becker, G.S., 1974. A theory of social interactions. J. Political Econ. 82 (6), 1063–1093. Bernard, H.R., 2011. Research Methods in Anthropology: Qualitative and Quantitative Approaches. Rowman Altamira, Lanham, MD. Blundo-Canto, G., Bax, V., Quintero, M., Cruz-Garcia, G.S., Groeneveld, R.A., PerezMarulanda, L., 2018. The different dimensions of livelihood impacts of payments for environmental services (PES) schemes: a systematic review. Ecol. Econ. 149, 160–183. Bottazzi, P., Wiik, E., Crespo, D., Jones, J.P.G., 2018. Payment for environmental “selfservice”: exploring the links between farmers’ motivation and additionality in a conservation incentive programme in the Bolivian Andes. Ecol. Econ. 150, 11–23. Bremer, L.L., Farley, K.A., Lopez-Carr, D., 2014. What factors influence participation in payment for ecosystem services programs? an evaluation of Ecuador’s SocioParamo program. Land Use Policy 36, 122–133. Bremer, L.L., Auerbach, D.A., Goldstein, J.H., Vogl, A.L., Chemie, D., Kroeger, T., Nelson, J.L., Benitez, S.P., Calvache, A., Guimaraes, J., Herron, C., Higgins, J., Klemz, C., Leon, J., Lozano, J.S., Moreno, P.H., Nunez, F., Veiga, F., Tiepolo, G., 2016. One size does not fit all: Natural infrastructure investments within the Latin American Water Funds Partnership. Ecosyst. Serv. 17, 217–236. Börner, J., Baylis, K., Corbera, E., Ezzine-de-Blas, D., Honey-Roses, J., Persson, U.M., Wunder, S., 2017. The effectiveness of payments for environmental services. World Dev. https://doi.org/10.1016/j.worlddev.2017.03.020. Deng, J., Sun, P., Zhao, F., Han, X., Yang, G., Feng, Y., 2016. Analysis of the ecological conservation behavior of farmers in payment for ecosystem service programs in ecoenvironmentally fragile areas using social psychology models. Sci. Total Environ. 550, 382–390. Engel, S., Pagiola, S., Wunder, S., 2008. Designing payments for environmental services in theory and practice: an overview of the issues. Ecol. Econ. 65 (4), 663–674. Figueroa, F., Caro-Borrero, A., Rvollo-Fernandex, D., Merino, L., Ameida-Lenero, L., Pare, L., Espinosa, D., Mazari-Hiriart, M., 2016. I like to conserve the forest, but I aslo like the cash”. Socioeconomic factors influencing the motivation to be engaged in the Mexican Payment for Environmnetal Services Programme. J. Forest Econ. 22, 36–51. Granovetter, M., 1985. Economic action and social structure: the problem of embeddedness. Am. J. Sociol. 91 (3), 481–510. Grillos, T., 2017. Economic vs non-material incentives for participation in an in-kind payments for ecosystem services program in Bolivia. Ecol. Econ. 131, 178–190. Grima, N., Singh, S.J., Smetschka, B., Ringhofer, L., 2016. Payment for ecosystem services (PES) in Latin America: analysing the performance of 40 case studies. Ecosyst. Serv.

41

Ecosystem Services 35 (2019) 32–42

K.W. Jones et al. on subjective wellbeing. World Dev. 105, 107–118. Rubin, D., 2006. Matched Sampling for Causal Effects. Cambridge University Press, New York. Samii, C., Lisiecki, M., Kulkami, P., Paler, L., Chavis, L., 2014. Effects of payment for environmental services (PES) on deforestation and poverty in low and middle income countries: a systematic review. Campbell System. Rev. 10 (11). Scullion, J., Thomas, C.W., Vogt, K.A., Perez-Maqueo, O., Logsdon, M.G., 2011. Evaluating the environmental impact of payments for ecosystem services in Coatepec (Mexico) using remote sensing and on-site interviews. Environ. Conserv. 38 (4), 426–434. Uchida, E., Xu, J., Xu, Z., Rozelle, S., 2007. Are the poor benefiting from China’s land conservation program? Environ. Dev. Econ. 12 (04), 593–620. Vaske, J.J., 2008. Survey Research and Analysis: Applications in Parks, Recreation and

Human Dimensions. Venture Publishing Inc, State College, PA. Woodhouse, E., Homewood, K.M., Beauchamp, E., Clements, T., McCabe, J.T., Wilkie, D., Milner-Gulland, E.J., 2015. Guiding principles for evaluating the impacts of conservation interventions on human wellbeing. Philos. Trans. R. Soc. B 370. Wunder, S., 2006. The efficiency of payments for environmental services in tropical conservation. Conserv. Biol. 21, 48–58. Wunder, S., 2015. Revisiting the concept of payments for environmental services. Ecol. Econ. 117, 234–243. Zanella, M.A., Schleyer, C., Speelman, S., 2014. Why do farmers join payments for ecosystem services (PES) schemes? An assessment of PES water schemes in Brazil. Ecol. Econ. 105, 166–176. Zbinden, S., Lee, D.R., 2005. Paying for environmental services: an analysis of participation in Costa Rica’s PSA program. World Dev. 33 (2), 255–272.

42