Cash for Conservation? The Environmental State,Public Payment for Ecosystem Services Programs,and Forest Loss in Low- and Middle-Income Nations

Abstract

The authors draw on environmental state theory to inform this study, examining the relationship between public payment for ecosystem services programs and forest loss from 2001 to 2015 in low- and middle-income nations. The authors analyze data for a sample of 79 low- and middle-income nations using a two-stage fractional regression model to address the potential issue of endogeneity on the public payment for ecosystem services program measures. The authors find support for environmental state theory that higher levels of payments to users as part of public payment for ecosystem services programs correspond to less forest loss in low- and middle-income nations. The authors also find that a composite variable, which measures key characteristics that may enhance the effectiveness of public payment for ecosystem services programs, has a similar relationship with forest loss in low- and middle-income nations. The authors conclude by discussing the theoretical, methodological, and policy implications of these findings.

Keywords

forest loss cross-national quantitative public payment for ecosystem services programs environmental sociology

According to the World Resources Institute’s (2022) Global Forest Watch, approximately 1.2 million square miles of forests were cleared between 2001 and 2015. However, more than 95 percent of the forest loss occurred in low- and middle-income nations (World Resources Institute 2022). This loss is equivalent 107 trillion metric tons of carbon dioxide emissions being released into the atmosphere, which contribute to climate change (World Resources Institute 2022). There are also localized environmental problems related to forest loss in low- and middle-income nations. They are biodiversity loss, flooding, soil erosion, and desertification (Seymour and Harris 2019). Furthermore, Seymour and Harris (2019) noted that forest loss also adversely affects people especially in low- and middle-income nations. The impacts involve lower agricultural yields, more infectious diseases spreading, and increasing human rights abuses especially among indigenous populations (Seymour and Harris 2019). With forest loss and its corresponding effects heavily concentrated in low- and middle-income nations, we focus our attention on them.

In this regard, the concept of the “environmental state” has been theorized (for recent reviews, see Rea and Frickel 2023 or Duit 2023). Duit, Feindt, and Meadowcroft (2016) argued that nations may address environmental issues in a few ways. To begin, a nation may set up specialized administrative units concerned with the natural environment. The units include ministries, parliamentary committees, or scientific advisory boards at the national, regional, and local level. A nation may also create and disseminate environmental knowledge by funding scientific research related to climate change, biodiversity, forests, water pollution, air pollution, or forest loss (Duit et al. (2016). Finally, Duit et al. noted that the most typical response involves implementing non-market-based policy instruments (i.e., command and control regulations) and market-based policy instruments.

Heine and Hayde (2023) argued that low- and middle-income nations are increasingly turning to market-based policy instruments or policies that seek to leverage market forces especially economic incentives to promote conservation. The most commonly used market-based policy instrument may be public payments for ecosystem services programs. By 2015, 25 public payment for ecosystem services programs had been implemented in low- and middle-income nations to address forest loss (Ezzine-de-Blas et al. 2016).

However, to our knowledge, we are not aware of any cross-national research that examines if public payment for ecosystem services programs may be related to forest loss in low- and middle-income nations. We are surprised by this for two reasons. First, meta-analyses of peer-reviewed articles on the topic have been carried out but have yielded contradictory findings. For example, Pattanayak, Wunder, and Ferraro (2010) and Alix-Garcia and Wolff (2014) found that such programs correspond with less forest loss. However, Samii et al. (2015) were slightly less optimistic in their review, noting, “The effect is modest . . . and seems to come with high levels of inefficiency,” which is quite a “troubling finding” (p. 30). Furthermore, Snilsveit et al. (2019) concluded, “Despite the hundreds of millions of dollars dedicated to payment for ecosystem service programs over the last decades . . . we are unable to determine with any certainty if they are worthwhile investments” (p. 67).

Second, data are available to assess the relationship between public payment for ecosystem services programs and forest loss in low- and middle-income nations. Ezzine-de-Blas et al. (2016) collected data for their meta-analysis on payment for ecosystem services programs that describe the extent of implementation and the characteristics that lead them to be successful in protecting forests. The authors have made the data available upon publication of their article. The data include information on public payment for ecosystem service programs that are searchable by country, start date, payment totals, and characteristics that tend to make a program effective in protecting forests (Ezzine-de-Blas et al. 2016).

We use these two points to begin our study but seek to move the cross-national research frontier on forest loss forward in a novel manner. In particular, we use environmental state theory to hypothesize that public payment for ecosystem services programs should be related to less forest loss in low- and middle-income nations. We begin by examining how the amount of money received by a provider as part of public payment for ecosystem services programs is related to forest loss in low- and middle-income nations. We then create and analyze a measure similar to Ezzine-de-Blas et al.’s (2016) indicator that captures compliance with six characteristics, which have been identified to improve the effectiveness of public payments for ecosystem services programs. We analyze the two measures using a two-stage fractional response regression model that addresses the limited nature of our dependent variable and the potential problem of endogeneity or the issue that public payments for ecosystem services programs are not randomly implemented in low- and middle-income nations (Wooldridge 2015).

We now turn to a discussion of environmental state theory and why it expects that public payments for ecosystem services programs are often used by low- and middle-income nations to address forest loss. We also discuss the reasons why they should correspond with less forest loss in low- and middle-income nations. We go on to describe our dependent variable, independent variables, and sample. We then discuss how a two-stage instrumental variable fractional regression model may help deal with endogeneity or nonrandom selection bias (Wooldridge 2015). We conclude by reviewing the findings along with the corresponding theoretical, methodological, and policy implications along with the directions for future research that follow from them.

Public Payments for Ecosystem Services Programs and Forest Loss in Theoretical Perspective

We note earlier that environmental state theory suggests the most common methods used by low- and middle-income nations to address forest loss are non-market-based and market-based policy instruments (Sommerer and Lim 2016). The most typical nonmarket policy instruments are direct and prescriptive methods that focus on setting specific requirements that companies or individuals must follow. In the forestry sector, such “command and control” measures include logging bans and export restrictions on timber. A typical market-based policy instrument seeks to leverage market forces and economic incentives to promote conservation. According to neoclassical economic theory, markets serve as the best way to set prices for goods and services (Heine and Hayde 2023). However, market failures are quite common, with environmental impacts of production, manufacturing, and distribution not being reflected in the prices of goods and services. When this occurs, negative externalities result (Heine and Hayde 2023). For example, a company that makes furniture may not consider the cost of carbon dioxide emissions that result from cutting down trees that serve as the raw material for their products. In such situations, a low- or middle-income nation may intervene to address negative externalities by implementing a public payment for ecosystem services program (Heine, Faure, and Dominioni 2020).

According to Wunder (2015), a public payment for ecosystem services program is a voluntary program funded, implemented, and managed by a low- or middle-income nation where users pay providers to manage their forests to maintain or enhance clean water, wildlife habitat, carbon storage, soil productivity, landscape beauty, or a combination of such services. It often does so by creating a formal market via legislation, which creates a demand for certain services by establishing a cap on the damage to a particular ecosystem or stimulating investment for a specific service (Ezzine-de-Blas et al. 2016). The users tend to be companies responsible for forest loss, which are seeking to reduce their impacts because of reputational problems or shareholder activism (Wunder 2015). The providers are individuals, communities, or companies that that agree to manage forests that they own or lease in a way that maintains or enhances the ecosystem services described previously (Engel, Pagiola, and Wunder 2008).

Although the process of implementing a public payment for ecosystem services program varies from country to country, Wunder (2015) described four characteristics that are typical of such a program and how they may enhance a program’s ability to address forest loss. We now turn to a discussion of the programs.

First, Wunder (2015) argued that it is necessary to identify what ecosystem services are to be protected and the locations that could supply such services. The demand for ecosystem services may be the result of pressure on a company by shareholders, nongovernmental organizations, social movements, or concerned citizens, to address its impacts on forests (Landell-Mills and Porras 2002). A low- or middle-income nation may identify an area that has many natural resources or is being threatened by mining, agriculture, or forestry (Sims and Alix-Garcia 2017).

Regardless of how a location is identified, it is essential for a low- or middle-income nation to define, measure, and determine, the value of any ecosystem services that a location possess (Wunder 2015). This is not a simple task and, typically, low- and middle-income nations assume a link between certain land uses and ecosystem service provisions (Wunder 2005). The assumption of a link tends to be the result of the limited financial capacity of governmental agencies to carry out the task and may undermine the effectiveness of a program (Wunder 2015). Ideally, a low- or middle-income nation determines a link between land use and a particular ecosystem service via locally proven scientific research (Wunder 2005).

Second, Wunder (2015) noted that low- and middle-income nations need to identify users and providers, who want to participate voluntarily in a public payment for ecosystem services program. A program should be voluntary because it ensures that payments are only made for actions that would not have occurred otherwise (James and Sills 2019). In other words, it addresses the issue that providers have been protecting ecosystem services on their land without any incentive (James and Sills 2019).

The identification of providers, who want to participate voluntarily, may seem like a fairly straight forward task. However, it is a complicated process because land rights are not often clearly defined in low- and middle-income nations (Alix-Garcia and Wolff 2014). As a result, it may be difficult to determine who is eligible to be a provider in such a program (Alix-Garcia and Wolff 2014). For example, a low- or middle-income nation may lease forests to a company or individual logging but not carbon sequestration or biodiversity protection or a low- or middle-income nation’s definitions of forest ownership, usage, or boundaries, may be different than customary rights within a nation (Walker et al. 2023). Thus, it is essential for a low- or middle-income nations to ensure ownership and use rights are defined, documented, and accepted, by all parties (Walker et al. 2023). The best way to build trust, dialogue, commitment, and, ultimately, voluntary participation by users is to employ a “free, prior, and informed consent framework” (Smith et al. 2019). There tend to be fewer issues with ensuring voluntary participation by users (Fripp 2014). As noted previously, many companies from high-income nations would like to offset the impacts of their activities and improve their public images. As such, they are looking for programs in which to participate (Wang and Wolf 2019).

Third, Wunder (2015) noted that a low- or middle-income nation negotiates a contract between users and providers. This stage addresses issues related to conditionality or the set of rules governing the market (Wunder 2005). A prominent topic involves setting the price for the services being offered (Vaissière et al. 2020). If a government pays private owners or managers of forests to maintain or enhance ecosystem services, then a user often has to accept a price set by legislation to that particular ecosystem (Stuart, Gunderson, and Petersen 2020). If a government creates a market, then the price tends to be set by supply and demand within regulatory parameters (i.e., price ceilings and floors) (Fripp 2014).

Not surprisingly, other factors go into setting price for providing ecosystem services. The most notable costs are related to monitoring and sanctioning (Fripp 2014). These requirements often need to be carried out by an independent, third party at multiple points in time (van der Ven and Cashore 2018). A public payment for ecosystem services program that do not have monitoring and sanctioning in place may lead to noncompliance by large numbers of providers (Ezzine-de-Blas et al. 2016). Put simply, monitoring and sanctioning are essential to deterring the free rider problem by providers (Wunder 2015).

Beyond setting the price, low- and middle-income nations work with providers to design management plans (Fripp 2014). The topics in the management plan tend to include mapping and clarifying the intricacies and complexities that exist between customary and civil law of a nation along with monitoring and evaluation programs that emphasize equity and fairness across the program (Wunder 2015). By addressing such issues, low- and middle-income nations can ensure that existing inequalities by race, ethnicity, gender, sexuality, ability, and class, are not reinforced (Walker et al. 2023).

There are other issues that need to be addressed at this stage. The most common include the integration of spatial targeting and differentiated payments (Ezzine-de-Blas et al. 2016). By including spatial targeting in a public payment for ecosystem services program, low- and middle-income nations focus on areas of high leverage or experiencing extensive forest loss, which has the potential to make a measurable impact (Ezzine-de-Blas et al. 2016). A differentiated payment scheme, where providers receive variable payments bases upon their holdings, may improve the cost efficiency than paying all providers that same (Ezzine-de-Blas et al. 2016). Toward this end, a public payment for ecosystem services program may have more money to invest (Ezzine-de-Blas et al. 2016).

Fourth, Wunder (2015) described how public payment for ecosystem services programs is implemented. If the preceding steps are followed, then a public payment for ecosystem services program may lead to less forest loss in a low- and middle-income nation. This is because providers receive payments to put natural resource management rules into place, which generate the ecosystem services identified to be protected or improved (Prokofieva 2016). On one hand, providers may receive funds not to clear forests, which, in turn, provide carbon sequestration or wildlife habitat (Wunder 2015). This practice is referred to as being “activity restricting” (Schomers and Matzdorf 2013). On the other hand, providers may receive payments to plant trees in previously cleared areas and engage in agriculture without clearing forests (Fripp 2014). These activities tend to be referred to as “asset building” (Schomers and Matzdorf 2013). Whether a program is activity restricting or asset building, a low- or middle-income nation needs transfer payments to providers directly (Muradian et al. 2010).

From the preceding discussion, we test the following two hypotheses:

Hypothesis 1: Higher levels of payments received by providers as part of a public payment for ecosystem services program will correspond with less forest loss in low- and middle-income nations.

Hypothesis 2: Higher scores on a composite measure including key characteristics of a public payment for ecosystem services measure will correspond with less forest loss in low- and middle-income nations.

Dependent Variable

Forest Loss Ratio

Recently, cross-national research has been published that uses satellite data on forest loss; see Sommer, Restivo, and Shandra (2024) and Harper, Sommer, and Shandra (2023) for examples. These data help eliminate measurement error found in previous estimates of forest loss from the United Nations Food and Agriculture Organization’s Global Forest Resource Assessments because of differences in collection methods varying from nation to nation (Hansen, Stehman, and Potapov 2010).

We calculate the forest loss ratio following Rudel (2017). First, we set the minimum tree cover canopy density level upon which to base the estimates. We set the minimum tree cover canopy density equal to 75 percent or greater to represent the loss of wet forests. The tree cover density for a nation represents the estimated percentage of a pixel taken from satellite imagery that is covered by tree canopy (World Resources Institute 2022). Second, we obtain the amount of the nation’s land area in hectares with the corresponding minimum tree cover canopy density (i.e., 75 percent). The data are available only for 2000. Third, we gather the number of hectares cleared between 2001 and 2015 in a nation. Fourth, we divide the total amount of hectares cleared between 2001 and 2015 by the total forest size in 2000 to compute the forest loss ratio.

Table 1 lists descriptive statistics and bivariate correlation matrix of forest loss analysis .

Table 1.

Descriptive Statistics and Bivariate Correlation Matrix of Forest Loss Analysis.

	Mean	SD	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
(1) Forest loss ratio	.162	.239	1.000
(2) Payments from public payments for ecosystem services programs	.118	.522	−.134	1.000
(3) Public payments for ecosystem services programs characteristics	.273	1.531	−.052	.781	1.000
(4) Total agricultural exports	2.051	1.100	.379	.072	−.075	1.000
(5) Agricultural land area	3.658	.694	.164	.014	.063	.033	1.000
(6) Democracy	−3.833	1.584	−.105	−.022	.135	−.146	−.238	1.000
(7) International nongovernmental organizations	3.001	.776	.043	.136	.226	−.184	.029	.246	1.000
(8) Gross domestic product per capita	6.741	1.048	−.238	.152	.218	−.396	−.168	.392	.214	1.000
(9) Economic growth rate	3.233	2.239	−.098	.048	−.024	.179	.016	−.279	−.355	−.244	1.000
(10) Total population growth rate	.208	.164	.414	−.076	−.063	.317	−.044	−.124	.078	−.432	−.089	1.000
(11) Rural population growth rate	.123	.224	.381	.001	−.170	.322	−.121	−.181	−.123	−.568	−.014	.762	1.000
(12) Urban population growth rate	.482	.368	.455	−.001	.041	.358	.132	−.198	.145	−.448	−.128	.812	.615	1.000

Independent Variables

Note that the model specifications and selection of independent variables are taken from Shandra, Restivo, and Sommer (2020) and Harper et al. (2023). The authors describe the theory linked to each independent variable. Note that we do not include each theory here. However, we do note cross-national research that has linked a given independent variable to forest loss in low- and middle-income nations.

Public Payment for Ecosystem Services Program’s Payment Amounts

We include the log of total payments provided to users per hectare of land as a result from a public payment for ecosystem services program in 2000. The data are measured in United States dollars using purchasing power parity exchange rates. We obtain the data from Ezzine-de-Blas et al.’s (2016) article. We expect that higher levels of payments received by providers should correspond with less forest loss in low- and middle-income nations.

Public Payments for Ecosystem Services Programs Characteristics

The second variable related to public payment for ecosystem services programs is an index that measures the degree to which a program has certain characteristics that enhance its effectiveness. Ezzine-de-Blas et al. (2016) identified several factors that help ensure public payment for ecosystem service programs decreases forest loss. These factors include: (1) how the link between land use and the ecosystem service being protected is identified, (2) voluntary participation, (3) monitoring compliance with natural resource management rules, (4) sanctioning for noncompliance, (5) the level of spatial targeting, (6) payment diversification, and (7) the degree to which payments are transferred directly to the providers. See Ezzine-de-Blas et al. (2016) for a complete discussion of each measure.

We compute this variable in the following way. First, we convert all of the variables to z scores, so they have the same metric. Second, we take the average of the standardized scores to compute the index. We expect that higher values on this measure should correspond with less forest loss in low- and middle-income nations.

Total Agricultural Exports

We include a country’s total agricultural exports as a percentage of total merchandise exports in 2000. We take the natural log of this variable to address its nonlinearity. The data may be obtained from the World Bank (2016). We expect higher levels of agricultural exports to correspond with more forest loss with companies clearing forests to increase the growing of crops for export (Jorgenson, Dick, and Shandra 2011).

Agricultural Land Area

We assess the impact of a nation’s agricultural sector by including the natural log of agricultural land as a percentage of a country’s total land area for 2000. The data may also be obtained from the World Bank (2016). We take the square root of this variable because of its nonlinearity. We hypothesize that low- and middle-income nations with a larger agricultural sector should have higher rates of forest loss as farmers grow crops for export or domestic consumption (Austin 2010).

Democracy

We use the average of Freedom House’s (2015) political rights and civil liberties scales to measure democracy in 2000 for each low- and middle-income nation The political rights scale reflects whether a nation is governed by democratically elected representatives and has fair, open, and inclusive elections. The civil liberties measure reflects whether a nation has freedom of the press, freedom of assembly, general personal freedom, freedom of private organizations, and freedom of private property. The variables are measured on a seven-point scale: free (1 and 2), partially free (3–5), and not free (6 and 7). We multiply the average of the scales by -1 so that higher scores equate to higher levels of democracy. We hypothesize that higher levels of democracy should correspond with less forest loss in low- and middle-income nations because of electoral accountability (Opoku and Sommer 2023) and concerned citizens being able to pressure the government for change by forming social movements and working with nongovernmental organizations (Marquart-Pyatt 2008).

International Nongovernmental Organizations

We include the number of international nongovernmental organizations working on environmental and animal rights issues in a nation for 2000. The data are collected by Smith and Wiest (2005) from the Yearbook of International Associations. Schofer and Hironaka (2005) found that higher levels of nongovernmental organizations are associated with lower rates of forest loss in low- and middle-income nations. This may be the case because nongovernmental organizations finance local conservation projects, support social movement activity, and shape the language of environmental laws (Jorgenson et al. 2011).

Gross Domestic Product

We include gross domestic product per capita for 2000 in our models. The data come from the World Bank (2016). We log this variable to correct for its skewed distribution. Burns, Kick, and Davis (2003) found that higher levels of gross domestic product per capita are associated with less forest loss. They argued that this is the case because wealthier nations tend to externalize their environmentally damaging activities by importing natural resources from elsewhere.

Economic Growth

We include the average annual economic growth rate from 2000 to 2015 from the World Bank (2016). We expect that economic growth rate to be associated with higher forest loss among other environmental issues (Clausen and York 2008). This may be the case because low- and middle-income nations, experiencing rapid economic growth, invest money in environmentally damaging activities like logging or export agriculture (Austin 2010).

Population Growth

It is also typical to control for the average annual percentage change in total population growth rate in cross-national models of forest loss (Allen and Barnes 1985). Rudel (1989) argued that the geometric growth in population outstrips arithmetic growth in the means of subsistence. As such, growing populations clear forests to raise crops to survive (Jorgenson and Burns 2007). From this observation, we hypothesize that population growth from 2000 to 2015 should correspond with increased forest loss.

Rural and Urban Population Growth

York, Rosa, and Dietz (2003) noted that it is important to “decompose” population dynamics in cross-national research. As such, we include the rural and urban population growth rates in our analysis. The variables are measured from 2000 to 2015. They may be obtained from the World Bank (2016). Jorgenson and Burns (2007) found that higher rates of rural population growth are related to increased forest loss, whereas higher rates of urban population growth are related to less forest loss. The authors argued that this is the case because expanding urban population relieve pressure on forests as migrants look for opportunities in manufacturing and service sectors found in cities. We expect a similar pattern of findings in our study.

Sample

The following 79 low- and middle-income nations according to the World Bank’s (2000) income classification scheme are included in the analysis following listwise deletion of missing data. We restrict our sample to low- and middle-income nations because 90 percent of forest loss is occurring there (World Resources Institute 2022). The sample includes Albania, Argentina, Armenia, Azerbaijan, Bangladesh, Belize, Benin, Bolivia, Botswana, Brazil, Bulgaria, Burundi, Cambodia, Cameroon, the Central African Republic, China, Colombia, Costa Rica, Cote d’Ivoire, Ecuador, El Salvador, Ethiopia, Equatorial Guinea, Eritrea, Fiji, Gabon, Gambia, Georgia, Ghana, Guatemala, Guinea, Guyana, Honduras, Hungary, India, Indonesia, Jamaica, Kazakhstan, Kenya, the Kyrgyz Republic, Macedonia, Madagascar, Malawi, Malaysia, Mali, Mexico, Moldova, Mongolia, Mozambique, Namibia, Nepal, Nicaragua, Nigeria, Pakistan, Panama, Papua New Guinea, Paraguay, Peru, the Philippines, Romania, Russia, Senegal, Sierra Leone, South Africa, Sudan, Suriname, Swaziland, Tajikistan, Tanzania, Thailand, Togo, Turkmenistan, Uganda, Ukraine, Vanuatu, Venezuela, Vietnam, Zambia, and Zimbabwe.

Statistical Model

It is common to use ordinary least squares regression in cross-national research on forest loss (e.g., Shandra, Rademacher, and Coburn 2016). However, it may be inappropriate to do so for a couple of reasons. First, the forest loss ratio is a proportion, so it ranges from 0 to 1. Toward this end, ordinary least squares regression may produce predictions that fall outside the range of the dependent variable and violate the of assumption of normality in the error terms (Long and Freese 2014).

Furthermore, ordinary least squares regression does not address the possible impact of endogeneity (Easterly 2006). In this instance, endogeneity on the public payment for ecosystem services program measures may lead to biased coefficients and inefficient tests of statistical significance (Wooldridge 2015). This occurs when one of the explanatory variables, on the right-hand side of the regression equation, is jointly determined with the left-hand side dependent variable to be explained (Wooldridge 2015). In such a case, an independent variable may be considered endogenous if the same processes that are related to forest loss in a country are also the same factors that are related to public payment for ecosystem services program implementation (Wooldridge 2015). In the end, ordinary least squares regression to estimate the effect of public payments for ecosystem services programs on forest loss will lead to biased estimates because this variable is not randomly assigned, and the regression equation is capturing the selection into creating a public payment for ecosystem services program as well as the effects of other independent variables on forest loss in one parameter (Wooldridge 2015). The estimate is also inconsistent, as the error term will be correlated with the endogenous predictor (Wooldridge 2015).

This appears to the case here. We find that the coefficients for the χ² test related to the Wald test of exogeneity are statistically significant in every model of Table 2. This indicates that the public payment for ecosystem services variables should be treated as endogenous.

Table 2.

Second-Stage Estimates of Instrumental Variable Fractional Regression Models.

	Model 2.1	Model 2.2	Model 2.3	Model 2.4
	Payments	Characteristics	Payments	Characteristics
Independent variables
Public payments for ecosystem services program measure, 2000	−1.307** (.485)	−.450** (.165)	−1.212* (.505)	−.448** (.171)
Total agricultural exports, 2000	.245* (.108)	.246* (.106)	.218* (.110)	.220* (.107)
Agricultural land area, 2000	.342** (.121)	.413** (.132)	.245* (.113)	.352** (.129)
Democracy, 2000	−.059 (.084)	.013 (.079)	−.046 (.083)	.028 (.077)
International nongovernmental organizations, 2000	.187 (.114)	.214 (.113)	.152 (.124)	.221 (.129)
Gross domestic product, 2000	.166 (.110)	.189 (.118	.142 (.086)	.208* (.095)
Economic growth rate, 2000–2015	−.014 (.042)	−.012 (.039)	−.005 (.043)	.001 (.039)
Total population growth, 2000–2015	1.552** (.547)	1.614** (.515)
Rural population growth rate, 2000–2015			.021 (.787)	.611 (.819)
Urban population growth rate, 2000–2015			.761 (.460)	.448 (.460)
Constant	−4.739*** (1.078)	−4.944*** (1.131)	−4.076*** (1.054)	−4.707*** (1.103)
Wald test of exogeneity	5.941*	5.558*	4.317*	4.842*
Anderson canonical correlation	9.400*	8.798*	11.822**	9.297**
Cragg-Donald F statistic	4.659	4.323	5.983	4.534
Sargan χ² test	1.491	.743	2.668	1.223
Bruesch-Pagan heteroskedasticity test	22.986*	18.173*	30.833*	17.147
Highest variance inflation factor	1.940	2.121	3.033	2.335
Mean variance inflation factor	1.448	1.555	1.810	1.760

Note: Regression coefficients are presented, with robust standard errors in parentheses.

p < .05, **p < .01, and ***p < .001 for a two-tailed test, with no directional hypothesis test being used for χ² statistics.

We attempt to address the endogeneity problem by using a two-stage instrumental variable fractional regression model (Wooldridge 2015). We carry out this analysis using Stata version 14 and the fracivp program created by Williams (2022).

The first stage model is estimated with ordinary least squares regression. In this instance, the endogenous variables—measures related to public payment for ecosystem services program—are regressed on two instrumental variables and the exogenous independent variables the second stage of the model. In the second stage of this model, the dependent variable is forest loss, which is bounded by 0 and 1 and, as such, is estimated by a quasi-likelihood or probit link function (Williams 2022). The model includes the predicted values of the endogenous variables from the first stage equation along with other exogenous independent variables.

The first instrumental variable that we use is the likelihood of being an ally of the United States in 2000. The data come from the Correlates of War Project (2015). We expect that low- and middle-income nations that are more closely aligned with the United States are more likely to put into place public payment for ecosystem services programs (Shorette 2014). We propose that this may be the case because the United States put one of the first programs into place and, as a result, has demonstrated its effectiveness and encourages low- and middle-income nations to follow suit (Shorette 2014). The second instrumental variable is the number of threatened bird species in a nation for 2000. The data come from the World Bank (2016). We hypothesize that low- and middle-income nations that have a large number of threatened bird species should be related to putting a public payment for ecosystem services program into place. We draw upon Rudel (2019) to argue that citizens press their governments to address environmental issues when confronted with such problems or what he terms “environmental shocks.”

Although it is important to identify the theoretical rationale for selecting an instrument, there are assumptions that must be met concerning the choice of instruments. The choice can be guided by various statistical tests (Wooldridge 2015). First, we must determine if an instrument is relevant or that it is correlated or statistically dependent with the endogenous variable but not the dependent variable (Wooldridge 2015). To test this assumption, we look at the coefficients for the Anderson canonical correlation. This χ² test reaches a level of significance in every model. Thus, we reject the null hypothesis and conclude that our instruments are relevant (Wooldridge 2015).

Second, we determine if the instruments are valid by testing to ensure that they are uncorrelated with the error term (Baum 2006). We consider Sargan χ² statistics for this purpose. The null hypothesis is statistical independence. The coefficients for this χ² test do not reach a level of significance in any model, thereby indicating the instrumental variables are not correlated with the error term (Baum 2006).

Third, we test to see if the instruments are weak or explain only a small amount of variation in the public payments for ecosystem services programs variables (Wooldridge 2015). This does not appear to be the case. We calculate a Cragg-Donald F statistic for each equation to reach this conclusion. We compare these values against Stock-Yogo critical values to determine if the instruments are weak (Wooldridge 2015). The coefficients for the F tests are greater than the Stock-Yogo critical values at a 15 percent to 20 percent bias level for every model (Wooldridge 2015).

Additional Regression Assumptions

We also ensure that we are not violating typical multivariate regression assumptions when using this model. We begin to do so by calculating the mean and highest variance inflation factor scores for each model. We report the values in Table 2. Tabachnick and Fidell (2012) noted that multicollinearity should not be a problem because the mean and highest variance inflation factor scores do not exceed a value of 2.5. This appears to be the case here. However, it is important to note that the highest variance inflation factor score equals 3 in model 2.3, but the mean variance inflation factor score remains in the acceptable range.

Second, we use Stata 14’s ladder and gladder commands to determine if a variable is normally distributed or needs to be transformed. The ladder command reports a χ² test for eight different transformations. The null hypothesis for the χ² test is that a specific transformation approximates normality (Tukey 1977). If the χ² statistic is statistically significant, then we reject the null hypothesis and conclude that the specified transformation does not approximate normality (Tukey 1977). We confirm the statistical tests by visually inspecting graphical distributions for each variable using the gladder command. We transform variables on the basis of the results and note any transformations (Tabachnick and Fidell 2012).

Third, we calculate standardized residuals to determine if outliers are a problem. We identify Cote d’Ivoire, Equatorial Guinea, Eritrea, and Namibia as multivariate outliers because their standardized residuals exceed an absolute value of 2.5 (Tabachnick and Fidell 2012). We remove them from the analysis and present the estimates produced without them.

Fourth, we calculate and report Breusch-Pagan tests for each model. The null hypothesis for this χ² test is that the error variances are homoscedastic or equally distributed (Tabachnick and Fidell 2012). The coefficients for the χ² statistics reach a level of significance, indicating potential problems with heteroscedasticity (Tabachnick and Fidell 2012). We report robust standard errors to address the issue.

Findings

In Table 2, we present the second-stage estimates of the two-stage instrumental variable fractional regression models. The first value presented is the regression coefficient and the second value, in parentheses, is the robust standard error. We report two-tailed significance tests at p < .05, p < .01, and p < .001 levels.

In every model, we include a measure related to public payments for ecosystem services, agricultural exports, agricultural land area, democracy, international nongovernmental organization presence, gross domestic product per capita, and economic growth. In odd-numbered models, we include total payments provided to users per hectare of land that is part of a public payment for ecosystem services program. In even-numbered models, we include the composite measure that includes characteristics of a public payment for ecosystem services program. In models 2.1 and 2.2, we include the total population growth. In models 2.3 and 2.4, we decompose total population into rural and urban population growth. We use this analytical strategy to build in “dimensions of variation” into the analysis, which allows us to look at how “cognate” but “distinct” indicators related to independent variables affect forest loss in low- and middle-income nations (Shandra, Ross, and London 2003).

We begin by considering the public payments for ecosystem services program variables. In odd-numbered models, we find that higher levels of payments to providers as part of public payment for ecosystem services program correspond with less forest in low- and middle-income nations. The coefficients for this variable are negative and significant in the odd-numbered models. In even-numbered models, we find a similar correlation with the composite measure that includes characteristics of public payment for ecosystem services programs that improve their effectiveness and forest loss. The coefficients for this measure are also negative and significant across Table 2.

We should also note the other factors that are related to forest loss in low- and middle-income nations. First, we find that agricultural exports correspond with increased forest loss. The coefficients for this variable are positive and significant in every model. Second, we find that higher levels of agricultural land area correspond with increased forest loss. The coefficients for this variable are positive and significant in Table 2. Third, we find that higher levels of total population growth are related to more forest loss in low- and middle-income nations. In models 2.1 and 2.2, the coefficients for this measure are positive and significant in Table 2.

We should also note the nonsignificant findings. We do not find that political characteristics predict significant variation in forest loss in our analysis. The coefficients for democracy and international nongovernmental organizations do not reach a level of statistical significance in Table 2. We also find that the coefficients that represent gross domestic product per capita and economic growth rate do not reach a level of statistical significance of statistical significance in the majority of models. Finally, we do not observe a statistically significant correlation between rural or urban population growth and forest loss in low- and middle-income nations.

Discussion and Conclusion

We move the cross-national research frontier on forest loss forward by testing how public payment for ecosystem services programs are related to forest loss in low- and middle-income nations. In doing so, we find support for environmental state theory that higher levels of payments as part of a public payment for ecosystem services program correspond with less forest loss in low- and middle-income nations. We also demonstrate that higher scores on variable measuring characteristics should be placed to make public payments for ecosystem services programs more effective and correspond with less forest loss. The coefficients for these two variables are negative and significant across a variety of alternative model specifications in Table 2.

We find other factors are related to forest loss in low- and middle-income nations. They include agricultural exports, agricultural land area, total population growth, and rural population growth. The coefficients for these variables are in line with theoretical expectations. They maintain their expected signs and significance levels across Table 2.

There are two theoretical implications that follow from this study. First, we begin by contributing to the long-running debate regarding if environmental reforms implemented by low- and middle-income nations are effective. On one hand, Mol and Buttel (2002) described how governments have put into place various environmental reforms that limit air pollution, water pollution, biodiversity loss, and forest loss. These include passing environmental laws, implementing market-based environmental instruments, and investing in environmentally friendly technologies. On the other hand, Schnaiberg, Pellow, and Weinberg (2002) argued that any reforms undertaken by governments are largely symbolic, which provide nations with “green” cover to support the pursuit of economic growth at the expense of the natural environment.

In this regard, we find that higher levels of money paid to providers as part of a public payment for ecosystem services programs are related to less forest loss in low- and middle-income nations. A similar finding emerges when we include our composite measure that measures characteristics that improve the effectiveness of public payment for ecosystem services programs. As such, we argue our findings suggest that low- and middle-income nations can address forest loss.

The second theoretical implication that follows from the first implication. We noted earlier that environmental state theory describes several ways that low- and middle-income nations may put into place to address forest loss. Duit et al. (2016) described several possible ways. They include setting up specialized administrative units concerned with the natural environment, creating environmental knowledge by funding scientific research, and putting regulations into place. In this study, we consider how a specific environmental regulation—a public payment for ecosystem services program—is related to forest loss in low- and middle-income nations. Toward this end, we refine environmental state theory by identifying a specific way that may low- and middle-income nations can address forest loss.

There are methodological implications of the study. First, we demonstrate the utility of using data on public payment for ecosystem services made available by Ezzine-de-Blas et al. (2016) to expand the cross-national research on forest loss. From these data, we examine not just the amount of cash paid to providers but also specific characteristics of public payment for ecosystem services programs that make them more effective. These include clearly defining the ecosystem services being protected, being completely voluntary, having monitoring and enforcement in place, and offering direct transfer of resources from users to providers.

The availability of such data provides social scientists with an opportunity to test hypotheses with more nuance than previously possible. It will serve social scientists well to consider if public payment for ecosystem service programs corresponds with less biodiversity loss, carbon dioxide emissions, and water scarcity among other environmental issues. There should also be research carried out that examines how different types of public payment for ecosystem service programs (i.e., activity restricting versus asset building) affect the natural environment.

We offer the following policy suggestions in an effort to address forest loss in low- and middle-income nations. Clearly, low- and middle-income nations should promote greater use and participation in public payment for ecosystem services programs (Prakash and Potoski 2012). However, it is essential that public payment for ecosystem service programs include specific characteristics if they are to be truly effective (Alix-Garcia and Wolff 2014). These characteristics are voluntary participation by users and providers, a clear definition of the ecosystem service being protected, monitoring to make sure providers are meeting established benchmarks, enforcement if they are not meeting benchmarks, and direct transfer of resources from users to providers (Wunder 2015).

We are not naive to think that this may be feasible for low- and middle-income nations that struggle to finance education, health, and poverty alleviation (Austin and Hof 2023). Thus, low- and middle-income nations may have to identify the revenue to support putting such a program into place. This revenue could come from eliminating subsidies for foreign companies in extractive sectors such as agriculture, mining, and forestry (World Bank 2023). The revenue could also come from expanding debt relief by participating in a bilateral debt-for-nature swap (Sommer, Restivo, and Shandra 2020) or engaging in a commercial debt-for-nature swap (Shandra et al. 2011).

There are limits to this study and directions for future research that follow. We analyze cross-sectional data because comparable satellite imagery data are not available over time. This is the result of changes in the collection methodology the World Resources Institute (2022). In particular, the denominator of our forest loss measure, total forest area, is available only for 2000 (World Resources Institute 2022). It would be informative to revisit and extend the analysis presented here when panel data become available. The availability of such data would allow the inclusion of country-specific and period-specific fixed effects, thereby controlling for invariant country characteristics of low- and middle-income nations (i.e., topography and climate) and time (i.e., economic recessions and expansions) (Jorgenson and Clark 2012).

Footnotes

ORCID iDs

Marion C. Harper

Jamie M. Sommer

Author Biographies

Marion C. Harper is a PhD student in the Department of Sociology at Stony Brook University. Her research focuses on how global governance and policies affect environmental outcomes cross-nationally.

Laura McKinney is an associate professor in the Department of Sociology at Tulane University. Her research focuses on environmental sociology, global political economy, climate change, gender, global and local sustainability, and rural and urban community development.

Michael Restivo is an associate professor and chair of the Department of Sociology at the State University of New York at Geneseo. His research focuses on global disparities in health, technology, and the environment.

John M. Shandra is a professor in the Department of Sociology at Stony Brook University. His research interests are in the areas of environmental sociology, globalization, and quantitative methodology, especially with regard to multinational corporations, international financial institutions, and repressive nations.

Jamie M. Sommer is an assistant professor in the Department of Sociology and Interdisciplinary Social Sciences at the University of South Florida. Her research interests are in the areas of governance and regulation in a global perspective, especially concerning the environment.

References

Alix-Garcia

Jennifer

Wolff

Hendrik

. 2014. “Payment for Ecosystem Services from Forests.” Annual Review of Resource Economics 6:361–80.

Allen

Julia C.

Barnes

Douglas F.

1985. “The Causes of Deforestation in Developing Countries.” Annals of the Association of American Geographers 75(2):163–84.

Austin

Kelly F.

2010. “Soybean Exports and Deforestation from a World-Systems Perspective: A Cross-National Investigation of Comparative Disadvantage.” Sociological Quarterly 51(3):511–36.

Austin

Kelly F.

Hof

Lukas

. 2023. “Growing Food, Feeding Disease: Primary Sector Specialization and Malaria Incidence in Less-Developed Countries.” Rural Sociology 88(2):486–511.

Baum

Christopher F.

2006. An Introduction to Modern Econometrics Using Stata. College Station, TX: Stata Press.

Burns

Thomas J.

Kick

Edward L.

Davis

Byron L.

2003. “Theorizing and Rethinking Linkages between the Natural Environment and the Modern World-System: Deforestation in the Late 20th Century.” Journal of World-Systems Research 9(2):357–90.

Clausen

Rebecca

York

Richard

. 2008. “Economic Growth and Marine Biodiversity: Influence of Human Social Structure on Decline of Marine Trophic Levels.” Conservation Biology 22(2):458–66.

Correlates of War Project. 2015. “Formal Alliances.” Retrieved May 20, 2025. http://www.correlatesofwar.org.

Duit

Andreas

. 2023. “The Environmental State.” Pp. 43–54 in Routledge Handbook of Environmental Policy, edited by Jörgens

Knill

Steinebach

London: Routledge.

10.

Duit

Andreas

Feindt

Peter H.

Meadowcroft

James

. 2016. “Greening Leviathan: The Rise of the Environmental State?” Environmental Politics 25(1):1–23.

11.

Easterly

William

. 2006. “Reliving the 1950s: The Big Push, Poverty Traps, and Takeoffs in Economic Development.” Journal of Economic Growth 11(4):289–318.

12.

Engel

Stefanie

Pagiola

Stefano

Wunder

Sven

. 2008. “Designing Payments for Environmental Services in Theory and Practice: An Overview of the Issues.” Ecological Economics 65(4):663–74.

13.

Ezzine-de-Blas

Driss

Wunder

Sven

Ruiz-Pérez

Manuel

Moreno-Sanchez

Rocio Del Pilar

. 2016. “Global Patterns in the Implementation of Payments for Environmental Services.” PLoS One 11(3):e0149847.

14.

Freedom House. 2015. “Freedom in the World.” New York: Freedom House.

15.

Fripp

Emily

. 2014. “Payments for Ecosystem Services (PES): A Practical Guide to Assessing the Feasibility of PES Projects.” Bogor, Indonesia: Center for International Forestry Research.

16.

Hansen

Matthew C.

Stehman

Stephen V.

Potapov

Peter V.

2010. “Quantification of Global Gross Forest Cover Loss.” Proceedings of the National Academy of Sciences 107(19):8650–55.

17.

Harper

Marion C.

Sommer

Jamie M.

Shandra

John M.

2023. “Government Revenue Composition and Forest Loss: A Cross-National Study of Low- and Middle-Income Nations.” Social Science Quarterly 104(6):1309–23.

18.

Heine

Dirk

Faure

Michael

Dominioni

Goran

. 2020. “The Polluter-Pays Principle in Climate Change Law: An Economic Appraisal.” Climate Law 10:94–115.

19.

Heine

Dirk

Hayde

Erin

. 2023. “Designing Fiscal Instruments for Sustainable Forests.” Washington, DC: The World Bank. Retrieved February 25, 2024. https://www.cif.org/sites/cif_enc/files/knowledge-documents/07_nationaltaxpolicyforcross-borderdeforestationproblems_web_cra.pdf.

20.

James

Natasha

Sills

Erin

. 2019. “Payments for Ecosystem Services: Program Design and Participation.” In Oxford Research Encyclopedia of Environmental Science. Oxford, UK: Oxford University Press.

21.

Jorgenson

Andrew K.

Burns

Thomas J.

2007. “Effects of Rural and Urban Population Dynamics and National Development on Deforestation in Less-Developed Countries, 1990–2000.” Sociological Inquiry 77(3):460–82.

22.

Jorgenson

Andrew K.

Clark

Brett

. 2012. “Are the Economy and the Environment Decoupling? A Comparative International Study, 1960–2005.” American Journal of Sociology 118(1):1–44.

23.

Jorgenson

Andrew K.

Dick

Christopher

Shandra

John M.

2011. “World Economy, World Society, and Environmental Harms in Less-Developed Countries.” Sociological Inquiry 81(1):53–87.

24.

Landell-Mills

Natasha

Porras

Ina T.

2002. “Silver Bullet or Fool’s Gold? A Global Review of Markets for Environmental Services and their Impact on the Poor.” London: International Institute for Environment and Development.

25.

Long

J. Scott

Freese

Jeremy

. 2014. Regression Models for Categorical Dependent Variables Using Stata. 3rd ed. College Station, TX: Stata Press.

26.

Marquart-Pyatt

Sandra T.

2008. “Are There Similar Sources of Environmental Concern? Comparing Industrialized Countries.” Social Science Quarterly 89(5):1312–35.

27.

Mol

Arthur P. J.

Buttel

Fredrick H.

2002. “The Environmental State under Pressure: An Introduction.” Pp. 1–11 in The Environmental State under Pressure, edited by Mol

A.P.J.

Buttel

F.H.

Leeds, UK: Emerald Group.

28.

Muradian

Roldan

Corbera

Esteve

Pascual

Unai

Kosoy

Nicolás

May

Peter H.

2010. “Reconciling Theory and Practice: An Alternative Conceptual Framework for Understanding Payments for Environmental Services.” Ecological Economics 69(6):1202–1208.

29.

Opoku

Alphonse

Sommer

Jamie M.

2023. “Can Democracy Reduce Forest Loss? A Cross-National Analysis.” World Development Sustainability 2:100073.

30.

Pattanayak

Subhrendu K.

Wunder

Sven

Ferraro

Paul J.

2010. “Show Me the Money: Do Payments Supply Environmental Services in Developing Countries?” Review of Environmental Economics and Policy 4(2):254–74.

31.

Prakash

Aseem

Potoski

Matthew

. 2012. “Voluntary Environmental Programs: A Comparative Perspective.” Journal of Policy Analysis and Management 31(1):123–38.

32.

Prokofieva

Irina

. 2016. “Payments for Ecosystem Services—The Case of Forests.” Current Forestry Reports 2(2):130–42.

33.

Rea

Christopher M.

Frickel

Scott

. 2023. “The Environmental State: Nature and the Politics of Environmental Protection.” Sociological Theory 41(3):255–81.

34.

Rudel

Thomas K.

1989. “Population, Development, and Tropical Deforestation: A Cross-National Study.” Rural Sociology 54(3):327–38.

35.

Rudel

Thomas K.

2017. “The Dynamics of Deforestation in the Wet and Dry Tropics: A Comparison with Policy Implications.” Forests 8(4):108.

36.

Rudel

Thomas K.

2019. “Shocks, States, and Societal Corporatism: A Shorter Path to Sustainability?” Journal of Environmental Studies and Sciences 9(4):429–36.

37.

Samii

Cyrus

Lisiecki

Matthew

Kulkarni

Parashar

Paler

Laura

Chavis

Larry

. 2015. “Decentralised Forest Management for Reducing Deforestation and Poverty in Low- and Middle-Income Countries: A Systematic Review.” London: International Initiative for Impact Evaluation.

38.

Schnaiberg

Allan

Pellow

David N.

Weinberg

Adam

. 2002. “The Treadmill of Production and the Environmental State.” Pp. 15–32 in The Environmental State under Pressure, edited by Mol

A.P.J.

Buttel

F.H.

Leeds, UK: Emerald Group.

39.

Schofer

Evan

Hironaka

Ann

. 2005. “The Effects of World Society on Environmental Protection Outcomes.” Social Forces 84(1):25–47.

40.

Schomers

Sarah

Matzdorf

Bettina

. 2013. “Payments for Ecosystem Services: A Review and Comparison of Developing and Industrialized Countries.” Ecosystem Services 6:16–30.

41.

Seymour

Frances

Harris

Nancy L.

2019. “Reducing Tropical Deforestation.” Science 365(6455):756–57.

42.

Shandra

John M.

Rademacher

Heidi

Coburn

Carolyn

. 2016. “The World Bank and Organized Hypocrisy? A Cross-National Analysis of Structural Adjustment and Forest Loss.” Environmental Sociology 2(2):192–207.

43.

Shandra

John M.

Restivo

Michael

Shircliff

Eric

London

Bruce

. 2011. “Do Commercial Debt-for-Nature Swaps Matter for Forests? A Cross-National Test of World Polity Theory.” Sociological Forum 26(2):381–410.

44.

Shandra

John M.

Restivo

Michael

Sommer

Jamie M.

2020. “Appetite for Destruction? China, Ecologically Unequal Exchange, and Forest Loss.” Rural Sociology 85(2):346–75.

45.

Shandra

John M.

Ross

Robert J. S.

London

Bruce

. 2003. “Global Capitalism and the Flow of Foreign Direct Investment to Non-core Nations, 1980–1996: A Quantitative, Cross-National Analysis.” International Journal of Comparative Sociology 44(3):199–238.

46.

Shorette

Kristen

. 2014. “Nongovernmental Regulation and Construction of Value in Global Markets: The Rise of Fair Trade, 1961–2006.” Sociological Perspectives 57(4):526–47.

47.

Sims

Katharine R. E.

Alix-Garcia

Jennifer M.

2017. “Parks versus PES: Evaluating Direct and Incentive-Based Land Conservation in Mexico.” Journal of Environmental Economics and Management 86(1):8–28.

48.

Smith

Tonya

Bulkan

Janette

Zerriffi

Hisham

Tansey

James

. 2019. “Indigenous Peoples, Local Communities, and Payments for Ecosystem Services.” Canadian Geographies 63(4):616–30.

49.

Smith

Jackie

Wiest

Dawn

. 2005. “The Uneven Geography of Global Civil Society: National and Global Influences on Transnational Association.” Social Forces 84(2):621–52.

50.

Snilsveit

Birte

Stevenson

Jennifer

Langer

Laurenz

Tannous

Natalie

Ravat

Zafeer

Nduku

Promise

Polanin

Joshua

, et al. 2019. “Incentives for Climate Mitigation in the Land Use Sector—The Effects of Payment for Environmental Services on Environmental and Socioeconomic Outcomes in Low- and Middle-Income Countries: A Mixed-Methods Systematic Review.” Campbell Systematic Reviews 15(3):e1045.

51.

Sommer

Jamie M.

Restivo

Michael

Shandra

John M.

2020. “India, Palm Oil, and Ecologically Unequal Exchange: A Cross-National Analysis of Forest Loss.” Sociological Perspectives 63(2):312–32.

52.

Sommer

Jamie Marie

Restivo

Michael

Shandra

John M.

2024. “Corruption and Palm Oil in a Cross-National Perspective: How India Contributes to Forest Loss in Peripheral Nations.” Journal of South Asian Development 19(1):83–111.

53.

Sommerer

Thomas

Lim

Sijeong

. 2016. “The Environmental State as a Model for the World? An Analysis of Policy Repertoires in 37 Countries.” Environmental Politics 25(1):92–115.

54.

Stuart

Diana

Gunderson

Ryan

Petersen

Brian

. 2020. Climate Change Solutions: Beyond the Capital-Climate Contradiction. Ann Arbor: University of Michigan Press.

55.

Tabachnick

Barbara G.

Fidell

Linda S.

2012. Using Multivariate Statistics. 6th ed. Boston: Pearson.

56.

Tukey

John

. 1977. Exploratory Data Analysis. Reading, MA: Pearson.

57.

Vaissière

Anne-Charlotte

Quétier

Fabien

Calvet

Coralie

Levrel

Harold

Wunder

Sven

. 2020. “Biodiversity Offsets and Payments for Environmental Services: Clarifying the Family Ties.” Ecological Economics 169:106428.

58.

van der Ven

Hamish

Cashore

Benjamin

. 2018. “Forest Certification: The Challenge of Measuring Impacts.” Current Opinion in Environmental Sustainability 32:104–111.

59.

Walker

Sarah

Alix-Garcia

Jennifer

Bartlett

Anne

Calder

Alice

. 2023. “Land Tenure Security and Deforestation: Experimental Evidence from Uganda.” Retrieved May 20, 2025. https://www.uvic.ca/socialsciences/economics/research/land_tenure_security_and_deforestation.pdf.

60.

Wang

Wolf

Steven A.

2019. “A Targeted Approach to Payments for Ecosystem Services.” Global Ecology and Conservation 17:e00577.

61.

Williams

Richard

. 2022. “Analyzing Proportions: Fractional Response and Zero One Inflated Beta Models.” Retrieved May 20, 2025. https://www3.nd.edu/~rwilliam/stats3/fractionalresponsemodels.pdf.

62.

Wooldridge

Jeffrey M.

2015. Introductory Econometrics: A Modern Approach. 6th ed., student ed. Boston: Cengage.

63.

World Bank. 2000. “Global Economic Prospects and the Developing Countries 2000.” Washington, DC: World Bank.

64.

World Bank. 2016. “World Development Indicators.” Washington, DC: World Bank.

65.

World Bank. 2023. “Transforming Trillions: Repurposing Subsidies for Climate Action and Economic Health.” World Bank. Retrieved February 26, 2024. https://www.worldbank.org/en/topic/climatechange/publication/detox-development.

66.

World Resources Institute. 2022. “Global Forest Watch.” Washington, DC: World Resources Institute.

67.

Wunder

Sven

. 2005. “Payments for Environmental Services: Some Nuts and Bolts.” Occasional Paper No 42. Bogor, Indonesia: Center for International Forestry Research.

68.

Wunder

Sven

. 2015. “Revisiting the Concept of Payments for Environmental Services.” Ecological Economics 117(C):234–43.

69.

York

Richard

Rosa

Eugene A.

Dietz

Thomas

. 2003. “Footprints on the Earth: The Environmental Consequences of Modernity.” American Sociological Review 68(2):279–300.