Ecological Economics 147 (2018) 333–342

Contents lists available at ScienceDirect

Ecological Economics
journal homepage: www.elsevier.com/locate/ecolecon

Analysis

Costs and Perceptions Conditioning Willingness to Accept Payments for T

Ecosystem Services in a Brazilian Case☆
⁎
Ronaldo Seroa da Mottaa, , Ramon Arigoni Ortizb
a
State University of Rio de Janeiro, UERJ, Av. Borges de Medeiros 3709 ap. 401, Lagoa, Rio de Janeiro 22470-001, RJ, Brazil
b
Rio de Janeiro Town Hall, Rua Leite Leal, 99 apto 16, Laranjeiras, Rio de Janeiro 22240-100, RJ, Brazil

A R T I C L E I N F O A B S T R A C T

JEL Classifications: This study analyzes the willingness of farmers to accept payments for ecosystem services in the Paraíba do Sul
Q51 River basin applying a contingent valuation methodology. Ecosystem services would be those resulting from
Keywords: forest conservation and regeneration and sustainable and innovative production practices. The results suggest a
Payment for ecosystem services regressive bias when some variables that capture the income effect positively affect participation in the program
Willingness to accept payment and acceptance of the payment offered. There is also evidence of adverse selection when acceptance of the
Contingent valuation amount offered is more sensitive to the service provision already being suited to the modes of production
adopted. The results, on the other hand, indicate that farmers' decisions to join the program depend not only on
their opportunity costs, but also on their perceptions about specific issues, such as their environmental
knowledge or awareness, inertia to change production modes, fear of additional monitoring, and level of un-
derstanding of the program. These results are important to support the program's outreach strategies and the
design of mechanisms for the selection of beneficiaries and pricing of payments.

1. Introduction agriculture (http://www.marcadagua.org.br).

This study analyzes the willingness of farmers to accept payments
for an ecosystem services program in the Paraíba do Sul river basin in
Brazil. This basin crosses the states of São Paulo, Minas Gerais, and Rio
de Janeiro, and covers an area of 56,500 km2 that includes the Paraíba
Paulista Valley and Fluminense regions and the Northwest Fluminense
and a major part of the Mata Mineira Zone.
Since colonial times the basin has been an agricultural center. With
the end of slavery and facing severe soil it experienced a decline in
agriculture with the expansion of dairy cattle farming. At the beginning
of the 20th century dynamic industrial activity begins and with the
heavy industrialization process in São Paulo and installation in the
basin, in 1946, of the Companhia Siderúrgica Nacional (CSN), the
National Steel-Making Company, the region becomes essential to the
economic integration of the states of Rio de Janeiro and São Paulo,
transforming the basin into one of the key communication and devel-
opment hubs for the region and for the country, thanks to the excep-
tional conditions that it offered –water supplies, sufficient electricity, a
consumer market, and the easy distribution of goods. Along its course
there is also a high urban density, dams for supplying water and gen-
erating hydroelectric power, mining sites, and substantial floodplain
With this long and intense human occupation, today the region
preserves < 20% of its original Atlantic Rainforest biome. This biome is
recognized as one of the five most important areas of biodiversity in the
world. Despite Regulation no. 11428, passed in 2006, prohibiting any
area of the Atlantic Rainforest being removed, the areas preserved still
face many external threats, such as illegal hunting and logging, and
accidental or natural fire.
In addition, there is forest fragmentation that, as well as the loss of
biodiversity, affects the region's water cycle, impacting water quality
and the recurrence of flooding and droughts. One of these more recent
consequences occurred at the beginning of 2010 when a flood caused by
raised waters damaged around 300 buildings in the city of São Luiz do
Paraitinga, displacing and leaving around 90% of the city's population
homeless and destroying a major part of the historic center.
Given the need to better conserve this remaining Atlantic Rainforest
and mitigate hydrological effects in the basin, state and federal gov-
ernments were engaged in developing a program of payments for eco-
system services (PES), which widens the current state initiatives and
specifically directs these towards an area of the basin in three states
located in four different municipalities, namely: São Luís do Paraitinga
and Santa Natividade in the state of São Paulo, Varre-Sai in the state of

☆
We thank two anonymous referees for valuable comments and suggestions all over the text. Errors remaining are, of course, only responsibility of the authors.
⁎
Corresponding author.
E-mail address: seroadamotta.ronaldo@gmail.com (R. Seroa da Motta).

https://doi.org/10.1016/j.ecolecon.2018.01.032
Received 26 March 2016; Received in revised form 15 May 2017; Accepted 22 January 2018
Available online 03 February 2018
0921-8009/ © 2018 Elsevier B.V. All rights reserved.
R. Seroa da Motta, R.A. Ortiz
Rio de Janeiro, and Muriaé in the state of Minas Gerais.1
The PES program under analysis intends to encourage the adoption
of a bundle of environmentally benign best practices that contribute to
the mitigation of hydrological effects in the basin. Such aim could be
achieved either by reducing degradation of current forestland or con-
verting current grazing area with reforestation or commercial forest
cultivation.
Since each activity bundle would result in different levels of activity
effort from farmers and, mainly, generating distinct level of service
provision, the PES program classified them as distinct services, namely:
Service 1 – Forest Conservation: activities to reduce forest and land
degradation, such as controlling fire, pests, invasion, hunting, palm
heart harvesting, and other sources of degradation and risks.
Service 2 – Forest Regeneration: activities to convert degraded
grazing land without any income alternative through the planting of
native trees.
Service 3 – Productive Conversion: activities to convert current
grazing land into rotational management and/or mixing agriculture and
grazing with forest cultivation.
Farmers could accept to participate in one or more of these services.
Those taking part will receive a six-monthly payment per hectare for a
period of three years.
Payment for ecosystem services (PES) has proved to be a widely-
used instrument for encouraging conservation and forest regeneration
in the world and in Brazil (Latacz-Lohmann & Hodge, 2003; UNEP,
2008; FAO, 2011 and Guedes & Seehusen, 2011).
Farmer participation in PES programs is thus conditioned by the
opportunity cost (direct and indirect) of the financial commitment
made by suppliers to offer these services. There is, however, extensive
literature that evaluates these experiments and points out that farmer
participation in PES is not only subject to the opportunity cost (direct
and indirect) for them There are also adverse selection problems when
participation tends to attract landholders who are already advanced in
the implementation of these services. Furthermore, regressive biases are
recorded when higher earning landholders, or those with large prop-
erties, and greater ability to mitigate risks, are more inclined to join the
program and/or are less demanding with regards to payment amount
(Alipizar, Blackman, & Pfaff, 2007, Wunder, Engel, & Pagiola, 2008,
Stefanie, Pagiola, & Wunder, 2008, Gomez-Baggethun, Groot, Lomasa,
& Montesa, 2010, Martin-Ortega, Ojea, & Roux, 2013 and
Grammatikopouloua, Ihoa, & Pouta, 2013).
In addition to opportunity cost, it is observed that the degree of
uncertainty regarding the program implementation, inertia in changing
production modes, and knowledge of the environmental impact of the
services also affect how farmers price joining these programs. (Grosjean
& Kontoleon, 2009; Cranford & Mourato, 2011 and Vijesh et al., 2013).
In the Brazilian case, Zanella, Schleyer, and Speelman (2014), by
carrying out an ex-post evaluation of the first PES program in Brazil for
water conservation practices, note that participation is correlated with
family labor intensity and access to information.
The same conditions appear in studies that analyzed program pro-
posals, identifying the degree of importance of these factors ex-ante and
the values that would attract the participation of farmers, with the aim
of contributing to the program's design and which could improve its
performance (Amigues, Boulatoff, Desaigues, Gauthier, & Keith, 2002,
Cooper, 2003, Ferraro, 2008, Buckley, Hynes, van Rensburg, & Doherty,
2009, Vatn, 2010; Cranford & Mourato, 2011and Leimona et al., 2015).
Stated preference techniques are usually applied in these cases of ex-
1
Together with the three states that share the Paraíba do Sul basin, and with the
support of the Global Environmental Facility (GEF) and the Inter-American Development
Bank (IADB). In São Paulo there is already a payment connected to forest conservation. In
Rio de Janeiro there is support for Private Nature Preservation Reserves (PNPR) and for
forest regeneration in Minas Gerais. This study benefited from valuable comments from
Simone Bauch, Helena Carrascosa, Fatima Casarin, Adriana Margutti and Carolina
Simone.
334
Ecological Economics 147 (2018) 333–342
ante evaluation. These techniques undertake surveys involving a sample
of potential participants of a PES program in a certain region in which
the program is presented, describing the activities to be remunerated
and questioning the interviewees about the factors that would affect
their participation and the amount that the interviewee would be pre-
pared to accept to join the program.
This study will thus identify, with a contingent valuation survey, the
economic and perception factors of agricultural landholders in the re-
gion, which affect participation in the PES program and, particularly,
the acceptance of the payment levels. Our contribution consists of the
first ex-ante evaluation of a PES program in Brazil, particularly with the
offer of three different service level provisions from distinct bundle of
environmentally benign best management practices, in which percep-
tion variables are included.
The next section presents the specification of the adopted model and
identification strategy. Next the survey is described and results follow.
Last section concludes.
2. The Model
Stated preference methods can be presented in two large groups: the
contingent valuation method (CVM), in which the person interviewed is
offered an amount to pay or accept a certain good or service provision
level; or the choice experiment (CE) in which interviewee reacts to pay
offers for attributes of the good or service. It is more difficult for the
interviewee to understand CE questions, despite it being the most
widely indicated technique when the attributes are relevant and un-
derstandable (Hoyos, 2010, and Hanley, Mourato, & Wright, 2001).
However, in this study we investigate the willingness to accept pay-
ments to each service provision and not attributes of them, therefore,
CVM is adopted.2
A contingent valuation survey is applied in a close-ended format.
First interviewees are asked if they would be willing to join the pro-
gram. If they would then they are asked if they would accept or not a
certain payment to delivery of each service. The analysis starts identi-
fying the factors that influence farmers in deciding to accept or not to
take part in the program and then the ones influencing acceptance of a
payment.
Formally, in this case, we take the rural landholder's expected utility
function as u (j, y; s), where j = 0 if the landholder maintains his/her
existing activities in an area of the land and j equals to 1 when he/she
accepts the service for the payment (PAY) offered to change these ac-
tivities. The y term is the landholder's income (rural and non-rural) and
s is the vector of features that affect the decision to join the program.
The landholder's expected utility in taking part in PES in exchange
for payment is given by u0 = u (1, y + PAY, s) and that for those who
do not take part is given by u1 = u (0, y; s). If u (j, y; s) is composed of
an observable component, v (j, y; s), and another non-observable one,
εj, so that u (j, y; s) = v,(j, y; s) + εj. The landholder will decide to join
if
v (1, y + PAY; s) + ε 0 > v (0, y; s) + ε1 (1)
or
v (1, y + PAY; s)–v (0, y; s) > ε1–ε 0 (2)
Assuming a distribution for the error term εj, the difference ε1 − ε0
would be distributed, for example, in Logit and Probit models. In other
words, the likelihood of the landholder accepting to take part in the PES
for the proposed payment would be when Δv equals the difference in
utility in Eq. (2) (Hanemann, 1984 and Maddala, 1983).
The validity tests for the results will be based on analysis of the
2
A CE approach to PES can be seen, for example, in Ruto and Garrod (2009) and Broch
and Vedel (2012) that investigate farmers´ stated preferences for attributes of afforesta-
tion contracts. That is, how much payment can be traded with contract terms.
R. Seroa da Motta, R.A. Ortiz Ecological Economics 147 (2018) 333–342

expected signs for the explanatory variables related to opportunity cost simulating a competitive tender with the threat of the program not
and perceptions, described as: taking place if the total payments exceed a certain pre-established (not
revealed) budget and if it does not achieve a participation level of 50%
(i) Economic: current income (agriculture and non-agriculture), cost of the area's agricultural region.
of production and rural credit; The need to analyze three services together adds another metho-
(ii) Productive: property size, family characteristics, type and use of dological point of interest relating to a potential sequence bias that is
land for activities, mode and intensity of production; affected by the order of the choice of services (Bateman & Langford,
(iii) Perception: environmental awareness, technical assistance, un- 1997 and Longo, Hoyos, & Markandya, 2015). Although in our case the
derstanding of the program, and motivation to change production occurrence of this bias can be minimized naturally since the services are
practices; different, in any case we adopt two procedures, namely: (i) the offered
(iv) Fixed effects: dummies for questionnaire type (X and Y, with in- payment value is the same for all three services in the pay acceptance
verted order of services) and for state or producer region; and question in each questionnaire and (ii) we create two different ques-
(v) Amount of payment offered: only for the payment acceptance tionnaires (versions X and Y), where the order in which the services are
function. presented is inverted.
The survey was undertaken for the 2012–2013 agricultural year. To
Property size, despite being an indicator of income expectation, capture language and cognitive elements for orientation in the design of
may, on the other hand, also capture diseconomies of scale that reflect the questionnaires, following the conventional procedures of the stated
the greater availability of unused areas, thus encouraging participation preferences methods, group meetings with farmers from the region
by reducing marginal opportunity cost. Similarly, size of forest area were carried out to adjust the language in the questionnaire to cognitive
should encourage joining. The current uses of the land resulting from barriers and tested in a pilot study with 30 open format interviews
investments, such as irrigation and forestry, should reduce these like- where the DAA values were elicited by the respondents. Based on these
lihoods. DAA elicited values, a scale of ten payment amounts was defined to be
However, a positive relationship is expected with variables related offered in the closed-ended format of the final survey.
to the modes of production associated with the services, such as low
productive intensity in forested areas, grazing rotation practices, and
3. The Survey
grazing land with trees and cultivation, which suggest some adverse
selection indicator, as the PES literature suggests, in which the program
The sample in the final survey consisted of 1079 questionnaires with
only attracts those who would be in a better position to carry out the
300, 306, and 473 from the municipalities of the states of Rio de
services. This would be even more regressive if there was also a positive
Janeiro (RJ), Minas Gerais (MG), and São Paulo (SP), respectively.
relationship with income indicators.
Tables 1 to 3 present some variables that characterize the properties
Perception variables that indicates inertia to changes in the modes
of production, fear of monitoring, concerns about loss of productivity,
Table 1
and mistrust with regards to the continuity of the program should be
Productive profile of properties (Rio de Janeiro).
negatively correlated with the likelihood of taking part in the program
and accepting the amount offered. Environmental awareness, under- Rio de Janeiro Obs Average value Standard Min Max
standing of the PES program, and the need for technical assistance and deviation
rural credit variables should be positively correlated.
Total occupants 299 3.48 1.60 1 9
Our questionnaire contained four parts, namely: (i) identification of Total property area (ha) 300 18.21 35.33 0 290.4
the landholder and his/her family; (ii) productive profile of his/her Forestland area (ha) 300 3.86 10.64 0 116.16
property; (iii) program presentation and questions relating to the Grazing ground area (ha) 300 8.28 19.80 0 174.24
landholder's perception; (iv) statement accepting or not to join the Cultivated area (ha) 300 5.64 12.80 0 145.2
Silviculture area (ha) 300 0.41 2.25 0 24.2
program; and then (v), if he/she joins, whether he/she accepts the Number of livestock 300 13.78 25.28 0 17
payment offered. In other words, the interviewee initially replies heads
whether he/she would take part in the program and if this is the case Number of livestock 299 6.28 15.00 0 120
he/she is further questioned. heads per capita
Coffee plantation area 300 8.99 70.70 0 1210
However, the most difficult question to address when applying the
(ha)
stated preferences method, and one which is widely discussed in lit- Sugarcane plantation 300 0.0077 0.0667 0 0.7
erature (Mitchell & Carson, 1989; Arrow et al., 1993, and Bateman area (ha)
et al., 2002), is the strategic bias that can over or underestimate the Eucalyptus plantation 300 0.043 0.334 0 4.84
stated preferences. In the case of the willingness to accept (WTA) the area (ha)
Number of fruit trees 300 40.65 198.29 0 2545
individual may focus of the payer's budgetary restriction and feel that Total cultivating cost (R 300 9713.62 26,773.87 0 350,000
his/her answer will be able to influence the way the good is provided, $/year)
but that he/she will not incur the costs of provision, and then reveal Total grazing feedstock 300 2074.95 6442.99 0 50,500
higher amounts than his/her true WTA (free rider effect). cost (R$/year)
Total income (R$/year) 222 34,750.91 36,150.13 0 200,000
To mitigate this bias, we used provision point mechanism techni-
Nonproductive income (R 300 10,429.04 16,369.86 0 134,400
ques that create a budgetary restriction for the payer (Murphy, Stevens, $/year)
& Weatherhead, 2005; Rose, Clark, Poe, Rondeau, & Schulze, 2002; Poe Total rural credit values 300 7656.50 14,543.15 0 96,000
et al., 2002 and Rondeau, Schulze, & Poe, 1999). In the PES literature (R$/year)
mentioned above, this mechanism is adopted based on an instrument in Family-based rural credit 300 5474.50 12,993.40 0 96,000
financing values (R
which it is established that if the sum of payments demanded exceeds a $/year)
pre-determined budget, the program will not take place, and in some Other variables Freq (%)
cases, this mechanism goes together with a competitive tender and an Father has always lived in the municipality 184 86.38
environmental target to be achieved based on the payments (Kline, Low use level of forestland 197 65.67
With grazing land in rotation 114 38.00
Alig, & Johnson, 2000; Vanslembrouck, Van Huylenbroeck, & Verbeke,
With enclosed grazing land 192 64.00
2002; Thomas & Blakemore, 2007; Cranford & Mourato, 2011, and With irrigated cultivation 101 33.67
Lindhjema & Mitani, 2012). In this study, we use a similar mechanism

335
R. Seroa da Motta, R.A. Ortiz Ecological Economics 147 (2018) 333–342

Table 2
Productive profile of properties (Minas Gerais).

Minas Gerais Obs. Average value Standard deviation Min Max

Total occupants 305 3.13 1.41 1 8

Total property area (ha) 306 50.85 445.64 0 6001.5
Forestland area (ha) 306 2.99 10.87 0 196.95
Grazing ground area (ha) 306 10.33 29.75 0 400
Cultivated area (ha) 306 36.87 428.35 0 5600
Silviculture area (ha) 306 0.55 2.13 0 28
Number of livestock heads Number of livestock heads 14.98 26.76 0 220
Number of livestock heads per capita Number of livestock heads per capita 5.58 10.03 0 73.33
Coffee plantation area (ha) 306 17.00 282.80 0 5000
Sugarcane plantation area (ha) 306 0.043 0.213 0 2
Eucalyptus plantation area (ha) 306 0.013 0.229 0 4
Number of fruit trees 306 28.80 188.02 0 3070
Total cultivating cost (R$/year) 306 5203.09 31,883.81 0 540,000
Total grazing feedstock cost (R$/year) 306 3260.29 8584.65 0 78,000
Total income (R$/year) 258 29,958.22 33,480.35 0 200,000
Nonproductive income (R$/year) 306 17,835.05 32,249.48 0 288,000
Total rural credit values (R$/year) 306 6199.58 28,809.49 0 315,000
Family-based rural credit financing values (R$/year) 306 3726.37 25,447.73 0 315,000
Other variables Freq (%)
Father has always lived in the municipality 200 85.11
Low use level of forestland 188 61.44
With grazing land in rotation 98 32.03
With enclosed grazing land 236 77.12
With irrigated cultivation 75 24.51

Table 3 As can be seen from Tables 1 to 3 below, the average property size is
Productive profile of properties (São Paulo). much larger in São Paulo, at around 130 ha, but with a high dispersion.
In Minas Gerais, the average area is 50 ha and in Rio de Janeiro, with
São Paulo Obs Average value Standard Min Max
deviation
the least dispersion, it is only 18 ha. While the average grazing plot, at
around 10 ha, is very similar in the three states, as well as, the number
Total occupants 473 2.91 1.69 1 9 of heads of livestock per capita, which is around six. In the three
Total property area (ha) 473 129.19 2020.94 0 43,560 states > 20% of occupants have already grazing land in rotation,
Forestland area (ha) 473 81.12 1362.82 0 29,040
Grazing ground area (ha) 473 11.70 26.59 0 239.58
and > 40% in Rio de Janeiro, 30% in Minas Gerais, and 20% in São
Cultivated area (ha) 473 35.26 668.21 0 14,520 Paulo stated that they had enclosed grazing land.
Silviculture area (ha) 473 0.97 7.04 0 99 As also indicated in Tables 1 to 3, the area of forest in Rio de Janeiro
Number of livestock 473 14.06 31.92 0 412 and Minas Gerais is around 4 ha. In São Paulo, due to the large dis-
heads
persion in the values measured, the average area of forest is 81 ha, that
Number of livestock 473 6.38 16.38 0 206
heads per capita is, a value that is much higher than the proportions of the productive
Coffee plantation area 473 0.512 11.128 0 242 areas. In the three states, 65% of the landholders said that they used
(ha) little of this forestland area.
Sugarcane plantation 473 0.052 0.418 0 7 Regarding income parameters, Tables 1 to 3 show that the average
area (ha)
Eucalyptus plantation 473 0 0 0 0
amounts of annual income per property are not low and are oddly
area (ha) higher in Rio de Janeiro and Minas Gerais, at around R$30,000, than
Number of fruit trees 473 47.43 283.13 0 3500 those of São Paulo, which are R$20,000, all of which have the least
Total cultivating cost (R 473 1327.24 10,082.65 1 210,000 dispersion. One possible explanation would be the much higher in-
$/year)
cidence of coffee plantations in Minas Gerais and Rio de Janeiro, since
Total grazing feedstock 473 2119.12 7930.62 0 78,000
cost (R$/year) for fruit trees, sugarcane, and eucalyptus, significant differences are not
Total income (R$/year) 335 19,756.88 22,035.87 0 200,000 observed. However, perhaps because of this, costs in these two states
Nonproductive income (R 473 13,673.80 18,000.16 0 168,000 are more than double those indicated in São Paulo.
$/year) Most of the financing needs come from family-based rural credit in
Total rural credit values 473 2680.55 12,343.39 0 153,000
(R$/year)
Rio de Janeiro and Minas Gerais but not in São Paulo, indicating that
Family-based rural credit 473 674.21 4738.05 0 80,000 there is different degree of insertion among regions into the commercial
financing values (R production chain.
$/year) To analyze the heterogeneity of the regional subsamples in a more
Other variables Freq (%)
statistically detailed way, Table 4 below presents how the productive
Father has always lived in the municipality 233 79.25
Low use level of forestland 318 67.23 profile of each state is similar or not according to the continuous de-
With grazing land in rotation 103 21.78 scriptive variables. For this, each pair of states was compared esti-
With enclosed grazing land 356 75.26 mating the t statistic to reject the null hypothesis that they are different.
With irrigated cultivation 69 14.59 The results in Table 4 clearly indicate that the heterogeneity be-
tween the subsamples is totally reflected in the opportunity cost in-
dicators, such as availability of family manpower (number of occu-
researched by region. It can be observed that the average number of
pants), economic results from the activities (revenues and costs), and
occupants is low, at around three for all of the states. There is little
productive investments (rotated or enclosed grazing land and irrigated
population mobility with > 80% of head of the families coming from
cultivation). The indicators relating to land use, such as property size,
the region.
livestock productivity, and areas dedicated to each activity are

336
R. Seroa da Motta, R.A. Ortiz Ecological Economics 147 (2018) 333–342

Table 4 perceptions answered with a 5 point agree/disagree Likert scale. It is

Heterogeneity of subsamples (t-test). observed in Table 5 that although the need to increase productivity is
scored about 40%, variables that would represent barriers to taking part
Variables RJ ≠ MG RJ ≠ SP SP ≠ MG
in the program or accepting payment, such as inertia (the desire not to
Total occupants 0.0045 0.0000 0.0614 change), fear of monitoring, lack of manpower, and mistrust, are be-
Total property area (ha) 0.2064 0.3420 0.5047 tween 20% and 30%. Perhaps, therefore, the perception that technical
Forestland area (ha) 0.3165 0.3266 0.3163
assistance and rural credit are important and reaches a frequency
Grazing ground area (ha) 0.3199 0.0554 0.5008
Cultivated area (ha) 0.2074 0.4431 0.9701 of > 60%.
Silviculture area (ha) 0.4493 0.1832 0.3065 Table 5 also indicate that about 80% of interviewees recognize the
Number of livestock heads 0.5717 0.8988 0.6761 importance of flood control. Familiarity with environmental issues that
Number of livestock heads per capita 0.3325 0.9376 0.2945 are relative to PES, issues of climate change and sustainability, and
Coffee plantation area (ha) 0.6374 0.0107 0.2103
legal conservation requirements are not low, presenting levels between
Sugarcane plantation area (ha) 0.0056 0.0689 0.7389
Eucalyptus plantation area (ha) 0.2312 0.0111 0.2140 50% and 80%. Understanding of the program was good for > 60% of
Number of fruit trees 0.4508 0.7173 0.3105 those interviewed and about 75% of them thought that the ques-
Total cultivating cost (R$/year) 0.0600 0.0000 0.0141 tionnaire was easy to answer.
Total grazing feedstock cost (R$/year) 0.0554 0.9355 0.0580
Such perceptions, as will be seen, will be crucial to analyze the
Total income (R$/year) 0.1325 0.0000 0.0000
Nonproductive income (R$/year) 0.0004 0.0116 0.0214 decision of farmers to join the PES program and accept payment for
Total rural credit values (R$/year) 0.4336 0.0000 0.0193 that.
Family-based rural credit financing values (R 0.2887 0.0000 0.0112
$/year)
Father has always lived in the municipality 0.7001 0.0381 0.0828 4. Factors Influencing Participation
Low use level of forestland 0.2803 0.6537 0.0982
With grazing land in rotation 0.1236 0.0000 0.0014 The interviewee initially answers whether he/she would take part in
With enclosed grazing land 0.0004 0.0008 0.5534
the program in general without indicating in which service and, if the
With irrigated cultivation 0.0130 0.0000 0.0005
answer is positive, is then questioned about each service separately and
Note: Pr(|T| > |t|) = P-value < 0,10 indicates inequality between the averages to a next if he/she accepts the payment amount offered. Table 6 indicates
degree of 90%. that 78% would accept taking part in the program, with small varia-
tions between regions.
statistically equal, except for the area of sugar cane, for the three However, as shown in Table 6, this level of participation varies
subsamples. These results will support the regional results for the fac- considerably where respondents are faced with the decision to parti-
tors influencing the decision to take part and accept the amount offered cipate in a specific service, with 50% of participation for conservation,
presented further on where the significances of the opportunity cost 46% for conversion, and only 26% for regeneration. As will be later
indicators appear in a more differentiated way for each subsample than discussed, the low level of participation in Service 2 – regeneration –
those for land use. may affect the analysis of the factors influencing the acceptance of
In Table 5 below, in turn, we analyze, for the combined sample of all payment for it.
three regions, the frequency of the variables that reflect respondents' In regional terms, Table 6 indicates that participation for services 1
and 2 in SP appears with a much higher percentage than the other two
Table 5 states, while for regeneration the percentage in the three states is al-
Frequency of perception variables. ways low but very close among the three regions.
In Table 7, using marginal effect estimators, the determinants of the
Variable % positive answers
decision to join the program are analyzed considering the total sample.
Member of cooperative 28.36 We use both the Probit and Logit models, which presented similar re-
Does not need additional income 15.4 sults. Here we report only those from the Probit one, which showed
Does not want to be monitored 16.7 better adjustment to the distribution of residuals. The marginal effects
Inertia to change current activities 23.6
Agrees that there is technical risk 16.9
estimation of a continuous independent variable (such as the averages
Does not like conservation or forest regeneration activities 5.1 of the predicted values) for categorical dependent variables, as is our
Inertia to adopt grazing land rotation 21.6 case of participation or acceptance of payment (Y-1) or non-participa-
Without family manpower for grazing land rotation 27.1 tion or acceptance (Y-0), shows how the likelihood of the dependent
Agrees with the importance of technical assistance 61.1
variable being true (Pr (Y = 1)) changes when the independent variable
Agrees with the importance of rural credit 63.4
Agrees with the need to increase productivity 41.5 (Xk) varies from a unit when all other variables in the model are
Is not confident he/she will be paid 27.2
Is not confident the program will take place 32.7 Table 6
Forest area already cut down 27.2 Sample of respondents.
Has understood the PES program 60.3
Believes PES is important for flood control 79.6 Sample of respondents All RJ MG SP
Owner of another property 40.2
Has already heard about the agroforestry system 45.9 Total of interviews 1079 300 306 473
Has already heard about untouchable forestland area legal 59.9 Number of those accepting to participate in any 843 269 265 309
requirements service
Has already heard about climate change 75.8 As % of the total sample 78,1 89,7 86,6 65,3
Has already heard about minimum forestland area legal 70.5 Number of those accepting to participate in service 1 - 543 179 177 187
requirements conservation
Has already heard about Rio + 20 54.5 As % of the total sample 50,3 59,7 57,8 39,5
Has already heard about ecosystem services 62.7 Number of those accepting to participate in service 2 - 277 90 72 115
Has already heard about sustainability 50.6 regeneration
Has already heard about PES 47.1 As % of the total sample 25,7 30,0 23,5 24,3
Member of a political party 25.5 Number of those accepting to participate in service 3 - 496 166 182 148
Low use level of forestland 65.2 conversion
Found the questionnaire easy 75.8 As % of the total sample 46,0 55,3 59,5 31,3

337
R. Seroa da Motta, R.A. Ortiz Ecological Economics 147 (2018) 333–342

Table 7 opportunity cost, such as, per capita area of grazing and cultivated land
Marginal effects of factors influencing participation. are statically significant and, as expected, negatively correlated to the
participation rate. To sum up, larger properties with a great proportion
Variables Coefficient Standard error
of forest area and less intensity of use have a greater incentive for
Education of the person responsible 0.008⁎⁎⁎ 0.002 taking part in the program.
Total occupants 0.032⁎⁎⁎ 0.007 There are also perception variables that turn out to be significant in
Little use of forested area 0.120⁎⁎⁎ 0.019
Table 7. A positive relationship appears with landholders having a
Has already heard about PES 0.068⁎⁎⁎ 0.018
Inertia to change current activities −0.146⁎⁎⁎ 0.028 higher level of education and them hearing about PES and having had a
Without family manpower for grazing land −0.127⁎⁎⁎ 0.026 greater understanding of the program. That is, familiarity with the
rotation subject, as expected, increases participation in the program what ex-
Agrees with the importance of technical assistance 0.077⁎⁎⁎ 0.019 plains technical assistance access motivating engagement.
Agrees with the importance of rural credit 0.064⁎⁎⁎ 0.020
As Table 7 shows, participation in the program is also positively
Has understood the PES program 0.088⁎⁎⁎ 0.019
Rio de Janeiro 0.097⁎⁎⁎ 0.025 correlated with the technical assistance and rural credit access to carry
Minas Gerais 0.116⁎⁎⁎ 0.021 out these services. However, participation falls when the landholder
Total property area 0.001⁎⁎ 0.000 thinks that he/she should not change his/her activities or if he/she
Percentage of forested area 0.134⁎⁎⁎ 0.050
thinks that rotated grazing land would occupy much more family labor.
Grazing area per capital −0.003⁎⁎ 0.001
Number of observations 0.008 To sum up, as expected, there exist potential regressive trends and
Pseudo R2 0.032 general participation in the program is negatively associated with op-
Log-Likelihood 0.120 portunity cost and availability of workforce. The understanding level of
Number of observations 1079 the program and access to technical assistance increase participation
⁎⁎⁎ but inertia to change production patterns acts in the opposite direction.
p < 0.01.
⁎⁎
p < 0.05.
⁎
p < 0.1. 5. Factors Influencing Acceptance of Payment

constant. In the cases of dummy variables this effect is measured esti-
mating the variation in likelihood when their values vary. The effects
comparison should be carried out only between variable categories.
As can be seen in Table 7, no direct income variable appears to be
significant. However, property size, that may indirectly reflect wealth,
shows a positive and statistically significant relationship that may be
indicating a regressive bias on the program assignment.
Table 7 also shows that the percentage of forestland area, with no
productive use and, consequently, reducing the opportunity cost, is
statistically significant and, as expected, generate a positive effect on
program participation on the whole. Conversely, indicators of high
The factors influencing acceptance of the payments offered are now
analyzed considering all of the states together as well as for each state
and each service individually. Regression results are presented also in
marginal effect estimators.
The regional aggregation offers a sample that can capture greater
variance; however, it covers a more varied range of properties with
their different productive, economic, and social characteristics. The
analysis for each state, inversely, allows for results that are specific to
the sample locations, although this may be hindered because of the
reduced sample size.
Similarly, we use the Probit and Logit models, which presented si-
milar results. Here we report only those from the Probit one, which

Table 8
Marginal effects of factors influencing acceptance of payment: all the states.

All regions Forest conservation Forest regeneration Productive conversion

Variables coef se coef se coef se

Payment amount for conservation 1.50e-04⁎⁎⁎ 1.73e-05

Monthly agricultural income 9.59e-06⁎⁎⁎ 3.11e-06
Total occupants 0.024⁎⁎⁎ 0.006
Silviculture area percentage −0.306⁎⁎⁎ 0.073
Average cost of cultivation 1.55e-05⁎⁎⁎ 4.55e-06
Percentage of reformed grazing land −0.112⁎⁎⁎ 0.020
Percentage of area with good cultivation production −0.113⁎⁎⁎ 0.021
Has already heard about climate change 0.213⁎⁎⁎ 0.023 0.146⁎⁎⁎ 0.034
Has already heard about sustainability −0.120⁎⁎⁎ 0.026 −0.113⁎⁎⁎ 0.036
Takes part in regeneration service 0.089⁎⁎⁎ 0.021
Takes part in conversion service 0.073⁎⁎⁎ 0.019
Payment amount for regeneration 7.88e-05⁎⁎ 3.20e-05
Forest area −1.46e-04⁎⁎⁎ 1.80e-05
Percentage of area under irrigated cultivation −0.305⁎⁎⁎ 0.063
Takes part in conservation service 0.152⁎⁎⁎ 0.033
Area under cultivation 0.003⁎⁎⁎ 0.001
Payment amount for conversion 1.32e-04⁎⁎⁎ 2.11e-05
Has already heard about PES −0.103⁎⁎⁎ 0.029
Agrees with the need to increase productivity
−0.071⁎⁎ 0.028
Number of observations 388 233 457
Pseudo R2 0.167 0.096 0.098
Log-Likelihood −150.06 −122.01 −200.60

Notes: coef = coefficient; se = standard error.

⁎⁎⁎
p < 0.01.
⁎⁎
p < 0.05.
⁎
p < 0.1.

338
R. Seroa da Motta, R.A. Ortiz
showed better adjustment to the distribution of residuals.
In the results described below we do not find any statistically sig-
nificant effect of the dummy from the type of questionnaire, which
indicates that the sequence in which the services are presented does not
influence the likelihood of the amount offered being accepted.
5.1. Aggregate
Table 8 presents the marginal effect of the acceptance of payment
offered functions for the aggregate of all of the three regions.
5.1.1. Payment – Conservation Service
As Table 8 shows, in the case of the conservation service, monthly
agricultural income is significant with a positive sign indicating a re-
gressive trend. Of the productive variables, negative effects on the
likelihood of acceptance come from the percentage of area designated
to silviculture activity, grazing land with rotation practices, and good
cultivation production, which increase opportunity cost, and therefore
reduce acceptance. Family size, indicating the availability of labor, also
increases acceptance positively. That is, for the opportunity cost in-
dicators, as expected, higher current and future income expectations
reduce the likelihood of acceptance. With regards to the perception
variables, having knowledge of climate issues leads to acceptance, but
knowledge of sustainable development does not.
Joining the other two services also increases the likelihood of ac-
ceptance, which may indicate that the landholder is evaluating a sum of
payments and, therefore, considering some economy of scope.
To sum up, for conversion practices, there is an indication of ad-
verse selection in acceptance of payment with regressive bias and evi-
dences of economies of scope with other services.
5.1.2. Payment – Regeneration Service
As indicated in Table 8, there is no significant economic variable to
explain the acceptance of payment for the regeneration service, but the
productive variables of having irrigated cultivation and the size of
forest cover appear to be significant and negative productive variables
that increase the opportunity cost of regeneration. Also, taking part in
the conservation service increases the acceptance of regeneration. To a
lesser extent, the payment amount is a variable with a positive effect.
To sum up, land uses that are more rigid to changes, such as those
with investment in irrigation and those with forest cover, reduce ac-
ceptance of payment for regeneration and the regeneration service have
economies of scope with conservation one.
5.1.3. Payment – Conversion Service
In the case of conversion practices, as shown in Table 8, no eco-
nomic or productive variable is observed to be significant. Of the per-
ception variables, having heard about sustainable development, as well
as PES, reduces the chances of the landholder accepting the payment
offered for the conversion service. With regards to climate change, the
effect is also positive, as in conservation. As expected, the need to in-
crease productivity in other areas of the property in order to be able to
accept payment for productive conversion practices also appears sig-
nificant and negative.
To sum up, acceptance of payments in the case of conversion
practices seems to be influenced by the landholder's familiarity with the
environmental issues of conversion, which may indicate a perception
that avoids the provision of valuable services (strategic bias), or in
finding that conversion is not financially profitable, primarily because
there is no significant relationship with opportunity cost variables, and
the income effect only appearing in the perception that there is the need
to increase productivity in the activities that are carried out in order to
compensate for the risk of conversion practices.
339
Ecological Economics 147 (2018) 333–342
5.2. Regional Effects
In the regional analysis, presented in Tables 9 to 11, we seek to
highlight the regional differences in the acceptance of payments offered
function.
5.2.1. Rio de Janeiro
The factors that influence acceptance of payment for conversion
practices for Rio de Janeiro from Table 9 include those with negative
effects from Table 8 for the aggregated sample with regards to the
variable that indicates knowing what PES is and having higher pro-
ductivity in the activities. And in the case of regeneration, possibly due
to the reduced sample, only the participation in conservation parameter
appears to be significant.
Unlike these two services, for conservation practices it is observed
that other variables appear to explain acceptance of the proposed
amount, with a strong marginal effect. With positive effects, as well as
income, little use of forest area, already having livestock enclosed, and
not wanting to engage in grazing land rotation all appear, again in-
dicating regressive and adverse selection bias. However, having another
property, which could be an indicator of income, appears with a strong
negative effect. Technical assistance access also increases acceptance.
For participating in conservation practices having heard about legal
forest conservation requirements, ecosystem services, and sustainability
reduces acceptance and, in particular, not wanting to be monitored has
negative marginal effect. In the productive conversion having heard
about PES reduces acceptance.
To sum up, acceptance of payment in Rio de Janeiro has a regressive
effect, with adverse selection bias, and reduces, as expected, with fear
of monitoring and sensitivity to perception about environmental issues
may indicate strategic bias for demanding a higher payment on socially
valuable services.
5.2.2. Minas Gerais
As Table 10 shows, in Minas Gerais acceptance of payment for
conservation is reduced by the negative effect of the landholder already
having knowledge of ecosystem services, while for conversion the re-
sults are similar of those obtained in the aggregated sample, with ad-
verse selection bias when opportunity cost indicators positively affect
acceptance.
On the other hand, acceptance of payment in the regeneration ser-
vice is only explained by the pay value. The acceptance of payment for
conversion practices is negatively related with livestock farming in-
tensity, or rather, as expected, it reduces when the grazing land is more
productive, that is, as opportunity cost increases. Knowledge about the
Conference Rio + 20 decreases the willingness to accept payment for
conversion service.
To sum up, the specificity of Minas Gerais would be that for con-
servation practices, as in Rio de Janeiro, would involve adverse selec-
tion, while for regeneration and conversion ones familiarity of en-
vironmental issues reduce acceptance.
5.2.3. São Paulo
In the case of São Paulo, as Table 11 shows, the factors influencing
regeneration practices are the same as those obtained in the aggregated
sample. In conservation, the number of occupants, other property, and
reformed grazing land variables repeat as significant and in equal ef-
fect, confirming the regressive pattern and adverse selection.
Knowledge of environmental issues does not influence the partici-
pation in the conservation service. But there is indication of economy of
scope with productive conversion.
Having heard about Rio + 20 has a negative effect on the pay ac-
ceptance of conversion practices. And in a much unexpected way, lack
of trust has a positive and expressive effect on the participation to the
conversion service.
To sum up, as well as indicating adverse selection and regressive
R. Seroa da Motta, R.A. Ortiz Ecological Economics 147 (2018) 333–342

Table 9
Marginal effects of factors influencing acceptance of payment: Rio de Janeiro.

Rio de Janeiro Forest conservation Forest regeneration Productive conversion

Variables coef se coef se coef se

Payment amount for conservation 2.42e-04⁎⁎⁎ 2.66e-05

Total annual income 2.04e-06⁎⁎⁎ 3.20e-07
Owner of another property −0.094⁎⁎⁎ 0.024
Little use of forest area 0.188⁎⁎⁎ 0.027
Percentage of reformed grazing land −0.213⁎⁎⁎ 0.021
Grazing land is enclosed 0.117⁎⁎⁎ 0.017
Percentage of area under irrigated cultivation −0.247⁎⁎⁎ 0.030
Has already heard about minimum forestland area legal requirements −0.158⁎⁎⁎ 0.034
Has already heard about untouchable forestland area legal requirements 0.220⁎⁎⁎ 0.027
Has already heard about sustainability −0.131⁎⁎⁎ 0.030
Has already heard about ecosystem services −0.156⁎⁎⁎ 0.049
Does not want to be monitored −0.200⁎⁎⁎ 0.052
Inertia in grazing land rotation 0.196⁎⁎⁎ 0.016
Percentage of area with good cultivation production −0.125⁎⁎⁎ 0.024
Agree with the importance of technical assistance 0.138⁎⁎⁎ 0.020
Agree with the importance of rural credit −0.106⁎⁎⁎ 0.034
Payment amount for regeneration 1.16e-04⁎⁎⁎ 4,23e-05
Forest area 0.071⁎⁎⁎ 0,017
Grazing land area −0.008⁎⁎⁎ 0,003
Payment amount for conversion 1,68e-04⁎⁎⁎ 3,30e-05
Has already heard about PES -0,191⁎⁎⁎ 0,054
Agrees with the need to increase productivity -0,150⁎⁎⁎ 0,046
Number of observations 133 90 166
Pseudo R2 0.588 0.162 0.169
Log-Likelihood −25.23 −43.76 −75.25

Notes: coef = coefficient; se = standard error.

⁎⁎⁎
p < 0.01.
⁎⁎
p < 0.05.
⁎
p < 0.1.

trends, the major specificity in São Paulo seems to be the weak influ- de Janeiro, and Muriaé in the state of Minas Gerais.
ence of familiarity with environmental issues. Our contribution consists of the first ex-ante evaluation of a PES
program in Brazil, particularly with three different services on offer,
and in which perception variables are included. Although our results
6. Conclusions and Recommendations
are in line with the existing PSE literature, they bring about important
questions to the program design.
This study has thus identified the economic and perception factors
In general, our results are aligned with expected sign and contents
of agricultural landholders in the region, which affect participation in
observed in the PES analytical literature. However, the discussion of
the program and, in particular, levels of payment when encouraged to
these results in the Program context can be very important since it al-
join a PES program.
lows, unlike other similar experiments, the design of the Program
This program was designed to pay for forest conservation, forest
benefits from an evaluation preceding its implementation.
regeneration, and productive conversion services in three states located
Firstly, our results indicated a low level of willingness to participate
in four different municipalities, namely: São Luís do Paraitinga and
in the program and such trend is negatively associated with the
Santa Natividade in the state of São Paulo, Varre-Sai in the state of Rio

Table 10
Marginal effects of factors influencing acceptance of payment: Minas Gerais.

Minas Gerais Forest conservation Forest regeneration Productive conversion

Variables coef se coef se coef se

⁎⁎⁎
Payment amount for conservation 1.62e-04 3.28e-05
Has already heard about PES −0.146⁎⁎⁎ 0.052
Area of forest 0.016⁎⁎⁎ 0.005
Area of grazing land per capita 2.64e-04⁎⁎⁎ 8.33e-05
Percentage of cultivated land with fruit trees 0.222⁎⁎⁎ 0.057
Payment amount for regeneration 3.18e-04⁎⁎⁎ 6.60e-05
Payment amount for conversion 1.33e-04⁎⁎⁎ 3.753e-05
Number of heads of livestock per capita −0.005⁎⁎⁎ 0.002
Has already heard about Rio + 20 −0.115⁎⁎ 0.053
Number of observations 133 39 146
Pseudo R2 0.205 0.110 0.122
Log-Likelihood −48.66 −22.09 −58.70

Notes: coef = coefficient; se = standard error.

⁎⁎⁎
p < 0.01.
⁎⁎
p < 0.05.
⁎
p < 0.1.

340
R. Seroa da Motta, R.A. Ortiz Ecological Economics 147 (2018) 333–342

Table 11
Marginal effect of factors influencing acceptance of payment: São Paulo.

São Paulo Forest conservation Forest regeneration Productive conversion

coef se coef se coef se

Payment amount for conservation 1.03e-04⁎⁎⁎ 2.79e-05

Total occupants 0.032⁎⁎⁎ 0.011
Owner of another property −0.112⁎⁎⁎ 0.043
Percentage of reformed grazing land −0.152⁎⁎⁎ 0.034
Has already heard about climate change 0.179⁎⁎⁎ 0.031
Area of grazing land 0.003⁎⁎ 0.001 −1.39e-03⁎⁎⁎ 4.10e-04
Payment amount for regeneration 2.05e-04 1.25e-04
Payment amount for conversion 7.21e-05⁎⁎ 2.98e-05
Has already heard about untouchable forestland area legal requirements 0.209⁎⁎⁎ 0.038
Has already heard about Rio + 20 −0.158⁎⁎ 0.066
Is not confident that he/she will be paid 0.147⁎⁎⁎ 0.019
Takes part in conservation service 0.097⁎⁎⁎ 0.028
Number of observations 182 32 145
Pseudo R2 0.159 0.032 0.183
Log-Likelihood −71.15 −13.43 −51.83

Notes: coef = coefficient; se = standard error.

⁎⁎⁎
p < 0.01.
⁎⁎
p < 0.05.
⁎
p < 0.1.

availability of family labor and the fear of changing production patterns
(inertia). Both issues are strongly related to the production pattern of
low-income farmers based on family structure that are financially weak
to face the risks of adopting non-conventional production patterns.
Such low levels of participation suggest that the design of the pro-
gram should improve communication mechanisms and be creative in
participation incentives that reduce the uncertainty level of engage-
ment, perhaps, by offering, along with payments, further technical as-
sistance coverage.
Secondly, as mentioned earlier, the evidence in the literature of
regressive bias in PES programs may also be present in this program
analyzed since results reported several positive effects of income-re-
lated variables as important influencing factors to participate in the
program. Therefore, those farmers with a lower income or opportunity
cost can be the least willing to join the program.
On the other hand, other indicators of opportunity cost could me-
liorate this regressive bias, such as: (i) the availability of labor, re-
flecting the number of occupants, tends to increase acceptance of
payment, and (ii) the expectation of returns from high investments in
production modes (sunk costs), such as property size, silviculture, ir-
rigation, and reformed grazing land, reduces the likelihood of accep-
tance.
Also important is the evidence of adverse selection when acceptance
of the amount offered is more sensitive to provision of the service al-
ready being suited to the adopted production modes, such as little use
of forest area and enclosed and rotated grazing land. In other words, as
already pointed out in the literature reviewed in this article, it may be
that respondents have stated a willingness to accept above the true
level, and that the program will attract those that are already offering
some of these services in some way. The results clearly confirm that the
likelihood of accepting the payment offered does not only depend on
the landholder opportunity cost, but is positively correlated with his/
her level of knowledge regarding environmental issues and negatively
with the fear of monitoring.
This possible selection bias requires incentives based on price dis-
crimination and, therefore, place great importance on the design of the
pricing mechanisms and on the beneficiary choice process. In addition
to that, their design must carefully articulate how monitoring and pe-
nalties could be enforced with great deal of participation and trans-
parency to mitigate non-compliance misjudgments and, perhaps, with
incentives for behavior adjustments.
Thirdly, the results also show that some economic factors influen-
cing on the payment acceptance may show up with opposite signals,
and services' synergy interactions as expected from the differences in
production profiles in each region. Regional results also reveal sig-
nificant variations regarding the perception factors related to environ-
mental knowledge and legal requirements and consequences of joining
the program in terms of monitoring and technical assistance.
In other words, the design of the program will have to explore these
regional aspects in their strategies for offering the services and may be
the case that regional designs and strategies must be needed instead of a
common approach to all municipalities and pricing instruments based
on productive vocation of each targeted region.
Finally, a strong synergy among services is also observed when it
comes to the decision to participate. Those joining to conservation
practices are less likely to participate to other services whereas parti-
cipation to regeneration increases willingness to join conversion ser-
vices. In the case of acceptance of payment offered, there is evidence of
economies of scope when accepting the payment for taking part in one
service increases the likelihood of the acceptance in others.
Synergy between services indicates that the design of strategies for
offering them will have to consider their negative and positive inter-
actions. To accomplish that, a careful analysis of the complementarity
of costs among the practices must be carried out and, ultimately, re-
vising the design of the services in distinct bundle of practices to ac-
commodate such features.
To sum up, the study results above all confirm the hypotheses and
conclusions from the literature. Interpretation of these results, however,
offers key suggestions for guiding the design of the program and its
instruments, and for thus avoiding undesirable biases that may com-
promise the efficiency of its implementation. They therefore highlight
that carrying out these economic analyses beforehand should form part
of the planning phase of any PES program.
References
Alipizar, F., Blackman, A., Pfaff, A., 2007. Payments for ecosystem services: why precision
and targeting matter. Resources 165, 20–22.
Amigues, J.P., Boulatoff, C., Desaigues, B., Gauthier, C., Keith, J.E., 2002. The benefits
and costs of riparian analysis habitat preservation: a willingness to accept/will-
ingness to pay contingent valuation approach. Ecol. Econ. 43 (1), 17–31.
Arrow, K., Solow, R., Portney, P.R., Leamer, E.E., Radner, R., Schuman, H., 1993. Report
of the NOAA Panel on contingent valuation. Fed. Regist. 58 (January, 10),
4601–4614.

341
R. Seroa da Motta, R.A. Ortiz
Bateman, Ian J., Langford, I.H., 1997. Budget constraint, temporal, and question-ordering
effects in contingent valuation studies. Environ. Plan. A 28, 1215–1228.
Bateman, I.J., Carson, R., Day, B., Hanemann, W.M., Hanley, N., Hett, T., Jones-Lee, M.,
Loomes, G., Mourato, S., Özdemiroolu, E., Pearce, D.W., Sugden, R., Swanson, J.,
2002. Economic Valuation with Stated Preference Techniques: A Manual. Edward
Elgar Publishing, Cheltenham.
Broch, W.S., Vedel, S.E., 2012. Using choice experiments to investigate the policy re-
levance of heterogeneity in farmer agri-environmental contract preferences. Environ.
Resour. Econ. 51, 561–581.
Buckley, C., Hynes, S., van Rensburg, T.M., Doherty, E., 2009. Walking in the Irish
countryside: landowner preferences and attitudes to improved public access provi-
sion. J. Environ. Plan. Manag. 52 (8), 1053–1070.
Cooper, J.C., 2003. A joint framework for analysis of agri-environmental payment pro-
grams. Am. J. Agric. Econ. 85 (4), 976–987.
Cranford, M., Mourato, S., 2011. Community conservation and a two-stage approach to
payments for ecosystem services. Ecol. Econ. 71, 89–98.
FAO, 2011. Payments for Ecosystem Services and Food Security. Food and Agriculture
Organization of the United Nations, Roma.
Ferraro, P.J., 2008. Asymmetric information and contract design for payments for en-
vironmental services. Ecol. Econ. 65, 810–821.
Gomez-Baggethun, E., Groot, R., Lomasa, P.R., Montesa, C., 2010. The history of eco-
system services in economic theory and practice: from early notions to markets and
payment schemes. Ecol. Econ. 69, 1209–1218.
Grammatikopouloua, I., Ihoa, A., Pouta, E., 2013. Willingness of farmers to participate in
agri-environmental auctions in Finland. Food Economics 9 (4), 215–230.
Grosjean, P., Kontoleon, A., 2009. How sustainable are sustainable development pro-
grams? The case of the sloping land conversion program in China. World Dev. 37 (1),
268–285.
Guedes, F.B., Seehusen, S., 2011. Pagamento por Serviços Ambientais na Mata Atlântica:
Lições aprendidas e desafios. Ministério do Meio Ambiente, Brasília.
Hanemann, W.M., 1984. Welfare evaluations in contingent valuation experiments with
discrete responses. Am. J. Agric. Econ. 66, 332–341.
Hanley, N., Mourato, S., Wright, R.E., 2001. Choice modelling approaches: a superior
alternative for environmental valuatioin? J. Econ. Surv. 15, 435–462.
Hoyos, D., 2010. The state of the art of environmental valuation with discrete choice
experiments. Ecol. Econ. 69, 1595–1603.
Kline, J.D., Alig, R.J., Johnson, R.L., 2000. Forest owner incentives to protect riparian
habitat. Ecol. Econ. 33 (1), 29–43.
Latacz-Lohmann, U., Hodge, I., 2003. European agri-environmental policy for the 21st
century. Aust. J. Agric. Resour. Econ. 47 (1), 123–139.
Leimona, B., Lusiana, B., Van Noordwijk, M., Mulyoutami, E., Ekadinata, A.,
Amaruzaman, S., 2015. Boundary work: knowledge co-production for negotiating
payment for watershed services in Indonesia. Ecosyst. Serv. 15, 45–62.
342
Ecological Economics 147 (2018) 333–342
Lindhjema, H., Mitani, Y., 2012. Forest owners' willingness to accept compensation for
voluntary conservation: a contingent valuation approach. J. For. Econ. 18 (4),
290–302.
Longo, A., Hoyos, D., Markandya, A., 2015. Sequence effects in the valuation of multiple
environmental programs using the contingent valuation method. Land Econ. 91 (1),
20–35.
Maddala, G.S., 1983. Limited-Dependent and Qualitative Variables in Econometrics.
Cambridge University Press, Cambridge.
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.
Mitchell, R.C., Carson, R.T., 1989. Using Surveys to Value Public Goods: The Contingent
Valuation Method. RFF Press, Washington, D.C.
Murphy, J., Stevens, T., Weatherhead, D., 2005. A meta-analysis of hypothetical bias in
contingent valuation. Environ. Resour. Econ. 30, 313–325.
Poe, G.L., et al., 2002. Provision point mechanisms and field validity tests of contingent
valuation. Environ. Resour. Econ. 23, 105–131.
Rondeau, D., Schulze, W.D., Poe, G.L., 1999. Voluntary revelation in the demand for
public goods using a provision point mechanism. J. Public Econ. 72, 455–470.
Rose, S., Clark, J., Poe, G.L., Rondeau, D., Schulze, W.D., 2002. Field and laboratory tests
of a provision point mechanism. Resour. Energy Econ. 24, 131–155.
Ruto, E.S.K., Garrod, G.D., 2009. Investigating farmers' preferences for the design of agri-
environment schemes: a choice experiment approach. J. Environ. Plan. Manag. 52
(5), 631–647.
Stefanie, E., Pagiola, S., Wunder, S., 2008. Designing payments for environmental services
in theory and practice: an overview of the issues. Ecol. Econ. 65, 663–674.
Thomas, R.H., Blakemore, F.B., 2007. Elements of a cost–benefit analysis for improving
salmonid spawning habitat in the river wye. J. Environ. Manag. 82 (4), 471–480.
UNEP, 2008. Payments for Ecosystem Services Getting Started: A Primer. Forest Trends,
The Katoomba Group, and UNEP.
Vanslembrouck, I., Van Huylenbroeck, G., Verbeke, W., 2002. Determinants of the will-
ingness of Belgian farmers to participate in agri-environmental measures. J. Agric.
Econ. 53 (3), 489–511.
Vatn, A., 2010. An institutional analysis of payments for environmental services. Ecol.
Econ. 69, 1245–1252.
Vijesh, V., Krishna, V.V., Drucker, A.G., Pascual, U., Raghu, P.T., King, E.D., 2013.
Estimating compensation payments for on-farm conservation of agricultural biodi-
versity in developing countries. Ecol. Econ. 87, 110–123.
Wunder, S., Engel, S., Pagiola, S., 2008. Taking stock: a comparative analysis of payments
for environmental services programs in developed and developing countries. Ecol.
Econ. 65, 834–852.
Zanella, M.A., Schleyer, C., Speelman, S., 2014. Why do farmers join payments for eco-
system services (PES) schemes? An assessment of PES water scheme participation in
Brazil. Ecol. Econ. 105, 166–176.