Contributed Papers

Costa Rica’s Payment for Environmental Services

Program: Intention, Implementation, and Impact
G. ARTURO SÁNCHEZ-AZOFEIFA,∗ ALEXANDER PFAFF,† JUAN ANDRES ROBALINO,†
AND JUDSON P. BOOMHOWER‡
∗
Earth Observation Systems Laboratory, Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton,
Alberta T6G 2E3, Canada, email [Link]@[Link]
†The Earth Institute, 2910 Broadway, Hogan Hall 104, Columbia University, New York, NY 10025, U.S.A.
‡Center for Conservation Biology, Department of Biological Sciences, 371 Serra Mall, Stanford University, Stanford,
CA 94305-5020, U.S.A.

Abstract: We evaluated the intention, implementation, and impact of Costa Rica’s program of payments
for environmental services (PSA), which was established in the late 1990s. Payments are given to private
landowners who own land in forest areas in recognition of the ecosystem services their land provides. To
characterize the distribution of PSA in Costa Rica, we combined remote sensing with geographic information
system databases and then used econometrics to explore the impacts of payments on deforestation. Payments
were distributed broadly across ecological and socioeconomic gradients, but the 1997–2000 deforestation
rate was not significantly lower in areas that received payments. Other successful Costa Rican conservation
policies, including those prior to the PSA program, may explain the current reduction in deforestation rates.
The PSA program is a major advance in the global institutionalization of ecosystem investments because few,
if any, other countries have such a conservation history and because much can be learned from Costa Rica’s
experiences.

Keywords: Costa Rica, deforestation trends, ecosystems services, payment for environmental services, PSA
El Programa de Pago de Servicios Ambientales de Costa Rica: Intención, Implementación e Impacto
Resumen: Evaluamos la intención, implementación e impacto del programa de pago de servicios ambi-
entales (PSA) de Costa Rica, que fue establecido al final de la década de 1990. Los pagos son otorgados a
propietarios privados en áreas boscosas como reconocimiento a los servicios ecosistémicos que proporcionan
sus tierras. Para caracterizar la distribución de PSA en Costa Rica, combinamos bases de datos de percepción
remota y de sistemas de información geográfica y posteriormente utilizamos econometrı́a para explorar los
impactos de los pagos sobre la deforestación. Los pagos fueron ampliamente distribuidos a lo largo de gradi-
entes ecológicos y socioeconómicos, pero la tasa de deforestación 1997–2000 no fue significativamente menor
en las áreas que recibieron pagos. Otras polı́ticas costarricenses de conservación exitosas, incluyendo algunas
previas al programa PSA, pueden explicar la reducción actual de tasas de deforestación. El programa PSA es
un avance importante en la institucionalización global de las inversiones en ecosistemas porque pocos, si
alguno, paı́ses tienen tal historia de conservación y porque se puede aprender mucho de las experiencias de
Costa Rica.

Palabras Clave: Costa Rica, pago de servicios ambientales, PSA, servicios ecosistémicos, tendencias de defor-
estación

Paper submitted March 21, 2006; revised manuscript accepted March 21, 2007.

1165
Conservation Biology Volume 21, No. 5, 1165–1173

C 2007 Society for Conservation Biology
DOI: 10.1111/j.1523-1739.2007.00751.x
 15231739, 2007, 5, Downloaded from [Link] by Universitatea Transilvania, Wiley Online Library on [15/05/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1166 Impacts of Ecosystem Services Payments Sánchez-Azofeifa et al.

Introduction (2001–today) and involves a comprehensive microtarget-

ing scheme and the provision of new ecosystem services
A new generation of conservation approaches is rapidly (e.g., drinking water) that were not part of the first phase.
emerging (Ferraro & Kiss 2002; Pagiola et al. 2002). They It is tempting to assign credit for the decrease in defor-
differ from traditional approaches in three critical and in- estation rates to the implementation of the PSA program.
terrelated ways: they emphasize human-dominated land- However, assigning causality to a farm-level program of
scapes, focus on ecosystem services, and utilize innova- this type requires a detailed analysis of deforestation over
tive finance mechanisms. To date, however, their imple- space and time and controls for other drivers of change
mentation has been on a small spatial scale. To apply the (e.g., international price of meat). Only by controlling
most promising initiatives worldwide, it is critical to un- for the effects of other factors (e.g., strong forestry leg-
derstand their intentions, designs, scopes, and limitations islation), can one conclude that the implementation of
(Ferraro 2001). Ecosystems provide services that include the PSA program and low rates of forest clearing are not
the pollination of crops, renewal of soil fertility, purifi- merely coincidental. Furthermore, how the PSA payments
cation of water, and stabilization of climate. For global were allocated needs to be taken into account.
services, such as the sequestration of carbon to aid in We examined the effect of the first phase (1997–2000)
climate stability, the origin of the service does not mat- of Costa Rica’s PSA program on ecological life zones, hy-
ter. Nevertheless, many services are supplied across lo- drological basins, buffer zones around protected areas,
cal and regional scales; thus, their delivery hinges on the planned biological corridors, and deforestation fronts. To
capacity of species and ecosystems to provide benefits test the hypothesis that the first generation of the PSA
precisely where humans are located. Therefore, success program has had an impact on the current low defor-
in conservation and many other aspects of human well- estation rates, we examined payment allocation relative
being is linked intimately to the management of highly to the future threat of clearing and land-use changes in
fragmented landscapes (Janzen 1998; McNeely & Scherr deforestation over space and time.
2002).
Incentives for maintaining the provision of ecosystem
services (Pagiola et al. 2002; Pagiola et al. 2004; Newburn
et al. 2005) include many components. Among them are Intent of Costa Rica’s PSA Program
regulatory systems of payments for ecosystem services,
such as those currently operating in countries such as Three laws form the framework within which Costa Rica
Australia, Costa Rica, and Mexico; market-based appro- established the PSA program. The 1995 Environment Law
aches to paying for ecosystem services, such as the emerg- 7554 mandates a “balanced and ecologically driven envi-
ing international carbon market; and mitigation banking ronment” for all. The 1996 Forestry Law 7575 mandates
approaches, such as those operating in the United States “rational use” of all natural resources and prohibits land-
in the context of the Clean Water Act and the Endangered cover change in forests. Finally, the 1998 Biodiversity Law
Species Act. Conservation programs under the European promotes the conservation and “rational use” of biodiver-
Common Agricultural Policy and the U.S. Farm Bill, in ad- sity resources.
dition to various financial-incentive schemes of The World Payments in the first phase were designed to address
Bank, are also important initiatives. relevant forest conservation failures from a legal and insti-
We explored the inception and initial impact of the tutional standpoint. The PSA program compensated forest
first generation (1997–2000) of an innovative conserva- landowners for value created by either planted or natu-
tion program established in Costa Rica: payments for en- ral forest on their land and recognized four services: (1)
vironmental services, or PSA (pagos por servicios am- greenhouse gas mitigation; (2) hydrological services; (3)
bientales). The PSA program in Costa Rica occurred in scenic value; and (4) biodiversity. The program did not
two phases. The first phase (1997–2000) coincided with attempt to measure all four services on a given parcel at
a significant drop in the national rate of deforestation once. An identically valued bundle of these services was
(1997–2000), relative to the 1986–1997 time period and assumed to be provided by each hectare of enrolled par-
the high rates of forest clearing that occurred from the cel. In the first phase enrollment was not based on parcel
1960s to the early 1980s (Sánchez-Azofeifa et al. 2001). size, and the policy was “first come, first served.” Fac-
Recently, there has been a net increase in forest cover, tors such as farm size, human capital, and household eco-
mostly due to land abandonment, although this process nomic level influenced participation in the program, and
is not sufficient to reverse the existing fragmentation of large landowners were disproportionately represented
the landscape or the increasing isolation of the country’s among participants at the national and regional levels (Mi-
national parks and biological reserves (Sánchez-Azofeifa randa et al. 2003; Zbinden & Lee 2005).
et al. 2003). The second phase of the PSA program re- The PSA program has to compete with other land-use
lates to the implementation of the Ecomarkets project returns. Average returns from PSA varied from US$22 to

Conservation Biology
Volume 21, No. 5, 2007
 15231739, 2007, 5, Downloaded from [Link] by Universitatea Transilvania, Wiley Online Library on [15/05/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Sánchez-Azofeifa et al. Impacts of Ecosystem Services Payments 1167

US$42/ha/year before fencing, tree planting, and certifi- plementation, however, rest with the Sistema Nacional de
cation costs. The main competing land use is cattle ranch- Areas de Conservacion (SINAC) and with the Ministerio
ing, which shows returns from US$8 to US$125, depend- del Ambiente y Energı́a (MINAE).
ing on location, land type, and ranching practices (Arroyo- The primary funding source for the original PSA pro-
Mora et al. 2005). One measure of cattle-ranching returns gram was a 15% consumer tax on fossil fuels estab-
is the cost of renting 1 ha of pasture. In Cordillera Cen- lished under the 1996 Forestry Law. Its Article 69 stated
tral, in the heart of Costa Rica, pasture rental ranges from that FONAFIFO was to receive one-third of the revenue.
US$20 to US$30/ha/year (Castro et al. 1998). The Ministry of Finance, however, rarely delivered that
amount, and in 2001 the legislature repealed Article 69
and adopted the Ley de Simplificación y Eficiencia Trib-
utaria, which assigns 3.5% of the tax revenue directly to
Implementation of the PSA the PSA program (Camacho & Reyes 2002). This provided
less money in theory, but increased actual transfers from
Three types of contracts were part of the first phase the Ministry of Finance (Camacho & Reyes 2002). As of
of the PSA program: forest conservation, reforestation, 2003, such tax revenues provided an average of $6.4 mil-
and sustainable forest management. Forest conservation lion/year to the PSA program (Pagiola et al. 2002).
contracts required land owners to protect existing (pri- Funding to the PSA program also comes from volun-
mary or secondary) forest for 5 years, with no land-cover tary contracts with private hydroelectric producers, who
change allowed. Reforestation contracts bound owners to reimburse FONAFIFO for payments given to individuals
plant trees on agricultural or other abandoned land and such as upstream landowners in watersheds. These pri-
to maintain that plantation for 15 years. Sustainable forest vate agreements have generated only about $100,000 to
management contracts (eliminated briefly in 2000) com- finance about 2,400 ha of PSA contracts. When fully im-
pensated landowners who prepared a “sustainable log- plemented, however, these agreements are expected to
ging plan” to conduct low-intensity logging while keeping provide about $600,000 annually and to cover close to
forest services intact. Just as in the reforestation contracts, 18,000 ha (Pagiola et al. 2002).
obligations for sustainable forest management contracts Carbon-abatement trading was expected to provide sig-
were for 15 years, although payments arrived during the nificant funding through sales of certified tradable offsets.
first 5 years. However, no significant market for carbon abatement has
Compensation varied across these types of contracts. emerged. The only sale has been to Norway, which con-
For conservation contracts, payment was US$210 (60,000 sisted of $2 million in 1997 for 200 million tons of carbon
colones)/ha in equal installments over 5 years. Reforesta- sequestration (Pagiola et al. 2002).
tion contracts paid US$537 (154,000 colones)/ha, with Funding was also provided by a World Bank loan and a
50% paid the first year, 20% the second year, and 10% Global Environmental Facility (GEF) grant through a pro-
over the following 3 years. The forest management con- gram called Ecomercados (a term used to define the sec-
tracts paid US$327 (94,000 colones)/ha, with the same ond phase of the PSA program after the year 2000). The
temporal structure as reforestation. Our conversion to World Bank/GEF loan for $32.6 million was designed to
U.S. dollars was based on the 1999 average exchange (287 support current PSA contracts. Of the total $8 million, $5
colones/U.S. dollar) rate provided by the Costa Rican Min- million was used for conservation contracts along the pro-
istry of Planning (MIDEPLAN). (Hereafter monetary units posed sites that will eventually form part of the Mesoamer-
are in U.S. dollars unless otherwise indicated.) ican Biological Corridor. The other $3 million was in-
Any PSA contract creates a legal easement that remains tended to increase human, administrative, and monitor-
with the property if it is sold. Owners transfer rights to the ing capacity in the various institutions associated with the
greenhouse-gas-mitigation potential of the parcel to the program, including FONAFIFO, SINAC, and MINAE (Ortiz
national government. Costa Rica can then sell these abate- & Kellenberg 2002).
ment units on any international market. Under the PSA
program rules, no individual can register <2 ha or >300
ha/year, although indigenous groups may register up to
600 ha/year. There is no area limit for coalitions that act Methods
through local nongovernmental organizations. Such orga-
nizations can function as intermediaries between small- Data on Payments, Forest Cover, and Dimensions of
Conservation Interest
holders and authorities to increase participation by those
who might not enroll. The data on the payment amounts for environmental ser-
The Fondo Nacional de Financiamiento Forestal (FON- vices to landowners were provided by FONAFIFO. The
AFIFO), a public forestry-financing agency created under information on contracts we analyzed is available in vec-
Forestry Law 7575 in 1996, administers the PSA program. tor format, including the spatial location of the farms that
The inspection responsibilities within the program’s im- participate in the different types of PSA contracts. We

Conservation Biology
Volume 21, No. 5, 2007
 15231739, 2007, 5, Downloaded from [Link] by Universitatea Transilvania, Wiley Online Library on [15/05/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1168 Impacts of Ecosystem Services Payments Sánchez-Azofeifa et al.

quantified forest cover and change based on a comprehen- ent life zones), its area was divided across those units as
sive Landsat Thematic Mapper data set produced for FON- a function of the total area falling within each category.
AFIFO by the University of Alberta’s Earth Observation To explore the effects of PSA contracts on changes in for-
Systems Laboratory (EOSL) and the School of Forestry est cover due to clearing, we used 5 × 5 km grid cells,
at the Costa Rica Technology Institute. The forest-cover with each cell representing a nested approach with a full
maps were produced for 1986, 1997, and 2000. The fol- resolution of 28.5 m and no resampling of the data.
lowing five land-cover categories were mapped: (1) forest We used forest and nonforest data (1986, 1997, and
(canopy closure > 80%), (2) 1986–1997 and 1997–2000 2000) from 2000 grid cells (total country area of 50,100
deforestation and reforestation, (3) mangroves, (4) non- km2 ) to calculate the 1986–1997 and 1997–2000 defor-
forest, and (5) cloud/water cover. The maps had a mini- estation rates. Using a GIS approach, we also overlaid PSA
mum mapping unit of 3.0 ha and were generated with the locations in the context of (1) proximity to national parks
same techniques implemented by the NASA Pathfinder and biological reserves, (2) slope, (3) aspect, (3) distances
Project (Skole & Tucker 1993; Sánchez-Azofeifa et al. to three major market locations (San Jose [capital city],
2001). Over 800 independent control sites were used Limon [Caribbean coast], and Puntarenas [Pacific coast],
for the validation of the 1997 and 2000 maps (Sánchez- and (4) life zones.
Azofeifa et al. 2001, 2003). Overall accuracy of the forest- Life zones were classified into three categories as a func-
cover map was 90% (Sánchez-Azofeifa et al. 2003). tion of their potential agricultural productivity: (1) “good”
We generated a digital elevation model (DEM) from 166 life zones, which included all humid areas (medium pre-
digitized topographic maps (1:50,000). Elevations every cipitation) with moderate temperature, (2) “medium” life
20 m and supplementary elevations every 10 m were used zone, which included very humid areas (high precipita-
to create a grid DEM, with a spatial scale of 28.5 m. We tion) in moderate to mountain elevations with moderate
used the DEM to produce drainage-basin boundaries and temperatures; and (3) “poor” life zones, which included
slope and aspect maps. all very humid areas with high temperatures, very dry, hot
A series of thematic GIS maps delineating different hy- areas, and rainy life zones (Pfaff & Sánchez-Azofeifa 2004).
drological, biological, and conservation land uses were We performed ordinary least square (OSL) regressions to
cross-referenced against the extent of national parks and explain the differences in deforestation rates across space
biological reserves. Those GIS maps were of (1) drainage and time based on PSA density (number of parcels per
basins important for hydropower production, drinking- hectare) and life zones, slopes, aspect, and distances to
water supply, and flood control (Sánchez-Azofeifa et al. major market locations.
2002a); (2) areas with potential for biological corridors Finally, we explored the hypothesis that implementa-
that may become part of the Mesoamerican Biological tion of the PSA program was based, in part, on expec-
Corridor efforts (known as GRUAS); (3) Holdridge life tations that payments for ecological services were made
zones (Holdridge 1967); and (4) political boundaries of in areas under greatest threat of forest conversion as sug-
the 13 conservation areas in Costa Rica. National park and gested by Pfaff and Sánchez-Azofeifa (2004). We examined
biological reserve areas falling in each one of the former this hypothesis by studying the proximity of the payments
GIS maps were eliminated because PSA payments are not to recent deforestation as an indicator of expected threat
allowed for public lands. (i.e., payments within 1 km of recent 1986–2000 defor-
From the point of view of water resource management, estation).
we placed all drainage basins into one of four categories:
(1) basins in which the management of water quality is
considered especially important (4 drainage basins), (2)
basins not important for water quality, (3) basins with Results
existing or planned dams for hydropower (8 drainage
Total Distribution of PSA Contracts
basins), and basins not planned for hydropower dams
(Sánchez-Azofeifa et al. 2002a). Finally, we explored the Around 300,000 ha of primary, secondary, or planted for-
spatial distribution of PSA in 1- and 5-km buffer areas est received funding in the first phase of the PSA program
around national parks, biological reserves, and 1997– through 2000. The mean project size was approximately
2000 deforestation fronts (Sánchez-Azofeifa et al. 2003; 102 ha. The largest project was 4025 ha. The stated size
Van Laake & Sánchez-Azofeifa 2004). limits were not fully enforced; 202 projects were over the
300-ha maximum and 60 contained less than the 2-ha min-
imum. From 1997 to 2000, the number of participants en-
Data Analysis
tering the program decreased (Fig. 1), probably because
To reveal the spatial distribution of PSA contracts, we funds were not delivered as expected (Camacho & Reyes
overlaid them with the GIS layers described above. For 2002).
these and other analyses, when a PSA polygon crossed a Payments for conservation alone were larger than the
border between two defined categories (e.g., two differ- sum of the payments made for reforestation and forest

Conservation Biology
Volume 21, No. 5, 2007
 15231739, 2007, 5, Downloaded from [Link] by Universitatea Transilvania, Wiley Online Library on [15/05/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Sánchez-Azofeifa et al. Impacts of Ecosystem Services Payments 1169

77.8%
100x103

80x103

73.1%
63.5%
60x103
Area (ha) under PSA

80.3%

52.9%
58.4%
20x103
Figure 1. Distribution of PSA
(payments for ecosystem
11.3%

20.5% services) in Costa Rica by type of

12.1%
payment (C, conservation; R,
10.9%

17.8%

21.1%
14.8%

25.9%
23.8%
19.7%
16%

reforestation; and M,
M management) between 1997 and
CR M CRM CR M CR CRM CR
0 2002. The number of people
Year: 1997 1998 1999 2000 2001 2002 involved in the program and the
Total Participants: (1531) (1021) (925) (501) (406) (329) total investment (in U.S. dollars)
Total Investment: 26,207,599 16,351,370 16,277,242 6,966,968 7,806,290 7,031,799 are presented for each year.

management (Fig. 1), but conservation contracts had the Spatial Allocation of PSA Contracts
lowest payments per unit area. Reforestation and man-
agement contracts generally held steady over the years, The PSA contracts were distributed broadly across ecolog-
whereas conservation payments fell (e.g., >$20 million ical life zones (Table 1). Most life zones had areas under
in 1997; almost $12 million in 1999; and <$4 million in PSA ranging from 4% to 8% of their total area. The max-
2001). imum allocation was achieved for Very Humid Tropical

Table 1. Distributions of PAS (payment for ecosystem services) on Costa Rican lands outside of protected areas between 1997 and 2000.a

Area or type of distribution

b
nonpublic nonpublic conservation reforestation management
parks (%) area (ha) area (%)b payment (%) payment (%) payment (%)

Holdridge life zones

Humid Lower Montane 0.0 23,925 0.8 0.1 0.7 0.0
Humid Premontane 13.0 553,550 4.4 3.6 0.8 0.0
Humid Tropical 1.1 1,068,499 7.2 4.1 2.8 0.4
Very Humid Montane 7.8 1,728 4.0 4.0 0.0 0.0
Very Humid Lower Montane 2.5 113,790 4.1 3.6 0.5 0.0
Very Humid Premontane 1.9 1,198,697 4.9 2.1 2.0 0.9
Very Humid Tropical 12.6 1,151,777 10.9 5.4 2.9 2.6
Pluvial Montane 54.6 127,785 3.7 3.0 0.7 0.0
Pluvial Lower Montane 45.5 348,464 7.8 7.0 0.5 0.2
Pluvial Premontane 24.5 373,038 6.9 6.7 0.2 0.0
Dry Tropical 3.3 141,544 6.3 4.9 1.4 0.0
Conservation areas
Amistad Atlantico 31.9 616,602 8.2 6.2 0.7 1.3
Amistad Pacifico 13.6 622,622 2.8 1.8 0.9 0.1
Arenal 5.1 917,380 10.4 2.8 4.8 2.7
Cordillera Central 10.3 569,277 8.0 5.2 1.8 1.0
Guanacaste 23.1 354,174 6.9 5.2 1.1 0.6
Osa 13.3 424,084 4.6 3.7 0.5 0.4
Pacific Central 1.2 545,339 5.1 2.8 2.3 0.0
Tempisque 3.4 760,534 7.4 6.1 1.3 0.0
Tortuguero 7.3 303,316 5.9 3.5 1.4 1.1
All of Costa Rica 11.3 4,534,350 7.0 4.1 2.0 0.9
a Distribution is presented across Holdridge life zones and conservation areas.
b Areas outside national parks and biological reserves.

Conservation Biology
Volume 21, No. 5, 2007
 15231739, 2007, 5, Downloaded from [Link] by Universitatea Transilvania, Wiley Online Library on [15/05/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1170 Impacts of Ecosystem Services Payments Sánchez-Azofeifa et al.

Table 2. Distribution of PSA (payments for ecosystem services) in Costa Rica from 1997 to 2000 for areas outside national parks and biological
reserves linked to conservation activities.

Area or type of distribution

nonpublica nonpublica conservation reforestation management
Conservation activity parks (%) area (ha) area (%) payment (%) payment (%) payment (%)

Urban water quality

San Jose’s & three othersb 6.7 1,102,459 3.6 2.3 1.1 0.2
all other basins 13.0 3,388,882 8.1 4.7 2.3 1.1
Hydropower interests
actual & planned damsc 13.4 1,597,442 5.1 2.6 1.8 0.7
all other basins 10.5 2,893,899 8.1 4.9 2.1 1.0
Near current protected areas
near (<5 km) 0.0 793,630 7.0 5.2 1.2 0.6
far (>5 km) 0.0 3,620,916 6.5 3.4 2.1 0.9
In planned protected areas
within GRUAS corridors 0.0 631,934 9.9 6.1 2.2 1.7
outside GRUAS corridors 0.0 3,787,737 6.3 1.9 0.8 3.6
a Nonpublic:areas outside national parks and biological reserves.
b Reventazon, Frio, Tarcoles, Terraba basins.
c Reventazon, Frio, Tarcoles, Terraba, Sixaola, Pacuare, San Carlos, Savegre basins.

Forest (10.9%) and the lowest allocation was observed Impact of the PSA Program on Deforestation
for the Humid Lower Montane (0.8%). Of the three types
Costa Rica experienced very low deforestation rates dur-
of incentives that were part of the first phase of the PSA
ing the study period. Deforestation rates were estimated
program, the conservation incentives dominated. Fewer
to be 0.06%/year and 0.03%/year for the 1986–1997 and
incentives for conservation were present in the Humid
1997–2000 time periods, respectively. Of the total distri-
Lower Montane life zone (0.1%) and more incentives were
bution of PSA payments in the country, only 7.7% were
in the Pluvial Lower Montane life zone (7%).
located within 1.0 km of all deforestation fronts. Conser-
There was a broad distribution of PSA contracts across
vation payments were higher (3.6%) than reforestation
the country’s nine conservation areas. These units mostly
(2.6%) and forest management (1.5%) payments in areas
had 4–8% of their total area under PSAs contracts, al-
close to deforestation fronts. A PSA payment was only
though Arenal (10.4%) and Amistad Pacifico (2.8%) dif-
slightly more likely to be near deforestation fronts (1 km)
fered from the other conservation areas. For these con-
than to be farther away from them. Our results also indi-
servation units, the spatial allocation across the types of
cate that there was no negative significant coefficient for
PSA contracts also varied.
the density of PSA payments (Table 3). The first genera-
Priorities for the allocation of PSA contracts as a func-
tion of the PSA program did not reduce deforestation rates
tion of hydrologic basins indicated two main trends (Ta-
or total deforestation in Costa Rica (Table 3, Fig. 2).
ble 2). First, from an urban quality point of view, PSA
contracts were allocated more often in basins with little
or no importance for drinking water (13% for all other
basins vs. 6.7% for San Jose’s and three other impor- Discussion
tant water-supply drainage basins). Second, actual and
planned basins with dams received more investment than Costa Rica’s PSA program has been a leader in the institu-
all other basins (13.4% vs. 10.0%, respectively). tionalization of ecosystem investments through the now
Our results (Table 2) also indicate that there was rel- popular idea of payments for ecosystem services. We doc-
atively little difference in the intensity of PSA contracts umented the spatial distribution of the first phrase of the
(number of parcels per unit area) relative to their prox- PSA program (1997–2000) through GIS overlays with var-
imity to conservation areas, and differed little among lo- ious indicators associated with conservation goals. The
cations (7.0% near [<5 km] conservation areas vs. 6.5% first phase of PSA contracts were distributed broadly
farther from [>5 km] conservation areas). Nevertheless, along these dimensions and lacked sharp asymmetries
the differences in intensity for areas within and outside that could indicate focused targeting. The sharpest asym-
planned biological corridors were larger (9.9% in the metry was for planned biological corridors. Furthermore,
GRUAS plans and 6.3% outside of the GRUAS plans). For all our econometric results suggest little, if any, impact of the
cases (hydrologic basins, conservation locations, planned level of PSA contracts on a given area’s rate of deforesta-
protected areas) conservation incentives dominated over tion. This has also been suggested by Hartshorn et al.
the reforestation and management incentives. (2005).

Conservation Biology
Volume 21, No. 5, 2007
 15231739, 2007, 5, Downloaded from [Link] by Universitatea Transilvania, Wiley Online Library on [15/05/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Sánchez-Azofeifa et al. Impacts of Ecosystem Services Payments 1171

Table 3. Results of regression analysis for Costa Rica’s 1997–2000 deforestation, PSA (payments for ecosystem services) (conservation,
reforestation, and management incentives [all] vs. conservation incentives alone), and selected control variables.a

I II III
conservation conservation conservation
all alone all alone all alone

Payments −0.0004 −0.004 0.001 −0.0005 −0.0003 −0.002

(0.92) (0.48) (0.83) (0.92) (0.95) (0.68)
Constant 0.004 0.004 0.016 0.016 0.011 0.011
(0.00) (0.00) (0.00) (0.00) (0.01) (0.01)
Slope −0.0003 −0.0003 −0.0002 −0.0002
(0.00) (0.00) (0.00) (0.00)
Good life zoneb −0.0002 −0.0002 0.0001 0.0001
(0.93) (0.94) (0.95) (0.94)
Poor life zoneb −0.004 −0.004 −0.002 −0.002
(0.02) (0.02) (0.36) (0.37)
Distance to San Jose −4e08 −4e08 4e09 4e09
(0.41) (0.40) (0.93) (0.93)
Distance to Limon −3e08 −3e08 −6e08 −6e08
(0.17) (0.17) (0.03) (0.03)
Distance to Puntarenas 3e08 3e08 8e09 8e09
(0.38) (0.38) (0.81) (0.81)
% clearedc 0.008 0.008
(0.00) (0.00)
Adjusted R2 0.00 0.00 0.04 0.04 0.04 0.04
n 2021 2021 1892 1892 1887 1887
a Allregressions were ordinary least squares for deforestation probabilities. Coefficient is reported; p value in parentheses. The dependent
variable was the deforestation rate during 1997–2000, measured at the level of the 5 × 5 km grid units of observation. The explanatory
variable of interest was the area receiving PSA payments, again measured by grid. We tried each of the two versions of the payments variable in
each of the three sets of columns. Columns II and III added more control variables (i.e., explanatory variables other than PSA).
b See Methods for definition of good and poor life zones.
c The fraction of the forest in a grid cell that was cleared before 1997.

From 1997 through 2000 little deforestation took place tion was consistently, positively, and significantly asso-
in Costa Rica (Figs. 2 & 3). Yet some explanatory effects ciated with deforestation. Distance to Limon was also
in the regressions were significant nonetheless (Table significantly associated with deforestation. These results
3). Slope in particular was consistently and negatively confirm many prior results in the literature (Velkamp
associated with deforestation, whereas prior deforesta- et al. 1992; Sánchez-Azofeifa et al. 2002b; Van Laake

1400
1.4

1200
1.2

1000
Total land on PSA (ha)

Deforestation Rate (%/yr)

1.0

800
0.8
600
0.6
400 Begining of the
0.4 first generation of
PSA in Costa Rica
200
0.2
0
0 50 100 150 200 250 0.0
Total Deforestation (ha) 1960-79 1979-86 1986-97 1997-00 2000-05
Deforestation Period
Figure 2. Relationship between the total area of farms
in the PSA (payments for ecosystem services) program Figure 3. Changes in deforestation rates between 1960
and the deforestation rate. The correlation coefficient and 2005. Arrow is implementation of the first PSA
between these two variables is 0.16. (payment for ecosystem services) program.

Conservation Biology
Volume 21, No. 5, 2007
 15231739, 2007, 5, Downloaded from [Link] by Universitatea Transilvania, Wiley Online Library on [15/05/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1172 Impacts of Ecosystem Services Payments Sánchez-Azofeifa et al.

& Sánchez-Azofeifa 2004; Schelhas & Sánchez-Azofeifa al. 2005) and some prior work in Costa Rica (Sierra &
2006, among others). With or without controlling for Russman 2006).
these other deforestation-driving variables, the level of Future work could also examine the effect of increased
PSA contracts in an area never generated significant sta- ecological targeting in the PSA program over time (e.g.,
tistical results concerning a reduction in the country’s post-2000). Perhaps future research could investigate
deforestation rate. whether the expected benefits from the PSA contracts
We believe the results from our regression analysis re- would rise even if impacts on the country’s deforestation
flect the significant impact of previous forest conserva- rates did not. Another important question to research is
tion policies in Costa Rica, including a 1997 legal restric- how impacts on deforestation could be increased. Fur-
tion on forest clearing (Law 7575, De Camino et al. 2000; ther targeting of deforestation threats in subsequent stud-
Hartshorn et al. 2005). All of the prior policies, including ies may help, but may require spatial variation in policies.
the creation of national parks and biological reserves and For example, research should address targeting and vari-
the 1997 law, have very effectively lowered deforestation. ations in payment levels ( Wilson et al. 2006).
Nevertheless, the roles of other changes, such as in mar- We believe Costa Rica’s pioneering effort opens the
ket prices, need to be considered when interpreting our door to further successful PSA programs. The evaluation
results. The success of these previous programs subse- of its first phase provides important conservation lessons
quently left the PSA program with little forest clearing to for developing countries interested in similar policies
prevent, thereby constraining the maximum possible im- (Newburn et al. 2005) and for new global policies.
pact of PSA payments, which turned out to be relatively
low as a result.
In the Costa Rican case PSA contracts may not have tar- Acknowledgments
geted deforestation pressure explicitly because this was
not the original purpose of the program. In fact, the PSA A Standard Research Grant from the Social Science and
contracts may have been targeted where there was a lack Humanities Research Council of Canada (SSHRC) to A.S.A
of deforestation pressure because the level of PSA con- funded this work. We thank the support and collabora-
tracts correlated negatively with the 1986–1997 forest tion of the Costa Rica National Forest Financing Fund
clearing. This could have resulted from a policy design (FONAFIFO). We thank three anonymous reviewers for
in which the PSA contracts were fixed across space (i.e., constructive and positive review comments and E. Main,
each of the locations in the country is assumed to provide who helped improve this manuscript. We also thank R.
the same services and is offered the same level of payment Chazdon, P. Ferraro, C. Leon, M. Guariguata, C. Orosz,
per hectare) and enrollment was voluntary. Targeting of M. Kalacska, and C. M. Rodrı́guez for constructive and
these kinds of lands could lead to those with unprofitable positive comments.
or low-profit land being the dominant participants in the
program. Literature Cited
The nonrandom location of the first phase of PSA con-
tracts is important to consider as well ( Wilson et al. 2006). Arroyo-Mora, J. P., G. A. Sánchez-Azofeifa, B. Rivard, J. C. Calvo, and D. H.
Given considerable agency expertise in Costa Rica, and Janzen. 2005. Dynamics in landscape structure and composition for
the Chorotega region, Costa Rica from 1960 to 2000. Agriculture,
good local data, the PSA program could have targeted Ecosystems and Environment 106:27–39.
threatened species so the impact could be greater than it Camacho, A., and V. Reyes. 2002. Pago por servicios ambientales. Oc-
initially appeared to be. Yet because land owners chose tavo informe sobre el estado de La Nacion en desarrollo humano
whether or not to participate, lower profitability lands sostenible. Consejo Nacional de Rectores, San José, Costa Rica.
could dominate the program and a large fraction of pay- Castro, R., F. Tattenbach, L. Gámez, and N. Olson. 1998. The Costa Rican
experience with market instruments to mitigate climate change and
ments could go to those who would have kept their lands conserve biodiversity. Fundacion para el Desarrollo de la Coordillera
in forest regardless of payments due to low opportunity Volcanica Central (FUNDECOR), San José, Costa Rica.
costs (Newburn et al. 2005; Sierra & Russman 2006). De Camino, R., O. Segura, L. G. Arias, and I. Perez. 2000. Costa Rica:
This self-selection for the PSA program could satisfy var- forest strategy and the evolution of land use. The World Bank, Wash-
ious policy goals, but it would not influence the country’s ington, D.C.
Ferraro, P. 2001. Global habitat protection: limitations of development
deforestation rate. Future work should examine more interventions and a role for conservation performance payments.
closely the extent to which such self-selection has in- Conservation Biology 15:990–1000.
fluenced the location of PSA contracts. Despite the fact Ferraro, P., and A. Kiss. 2002. Direct payments to conserve biodiversity.
that Costa Rica’s effective conservation history is some- Science 298:1718–1719.
what unique and that in many countries deforestation may Hartshorn, G., P. Ferraro, B. Spergel, and E. Sills. 2005. Evaluation of the
World Bank—GEF Ecomarkets project in Costa Rica. North Carolina
be higher, this policy design consideration is an impor- State University, Durham.
tant point that is relevant elsewhere. It is also consistent Holdrighe, L. 1967. Life zone ecology. Tropical Science Center, San José,
with work on policy location in Mexico (Muñoz-Piña et Costa Rica.

Conservation Biology
Volume 21, No. 5, 2007
 15231739, 2007, 5, Downloaded from [Link] by Universitatea Transilvania, Wiley Online Library on [15/05/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Sánchez-Azofeifa et al. Impacts of Ecosystem Services Payments 1173

Janzen, D. H. 1998. Gardenification of wildland nature and the human Sánchez-Azofeifa, G. A., R. C. Harriss, L. Storrier, and T. De-Camino-
footprint. Science 279:1312–1313. Beck. 2002a. Water resources and regional land cover change in
Miranda, M., I. Porras, and M. L. Montero. 2003. The social impacts of Costa Rica: impacts and economics. International Journal of Water
payments for environmental services in Costa Rica: a quantitative Resources Development 18:109–424.
field survey and analysis of the Virilla watershed. Environmental Sánchez-Azofeifa, G. A., B. Rivard, J. C. Calvo, and I. Moorthy. 2002b.
Economics Programme, London. Dynamics of tropical deforestation around national parks: remote
McNeely, J. A., and S. J. Scherr. 2002. Ecoagriculture: strategies to feed sensing of forest change on the Osa Peninsula of Costa Rica. Moun-
the world and save wild biodiversity. Island Press, Washington, D.C. tain Research and Development 22:352–358.
Newburn, S. Reed, P. Berck, and A. Merenlender. 2005. Economics and Sánchez-Azofeifa, G. A., G. Daily, A. Pfaff, and C. Busch. 2003. Integrity
land use change in prioritizing land conservation. Conservation Bi- and isolation of Costa Rica’s national parks and biological reserves:
ology 19:1411–1420. examining the dynamics of land cover change. Biological Conserva-
Muñoz-Piña, C., A. Guevara, J. M. Torres, and J. Braña. 2005. Paying for tion 109:123–135.
environmental services of Mexico’s forests: analysis, negotiations Schelhas, J., and G. A. Sánchez-Azofeifa. 2006. Post-frontier forest change
and results. Instituto Nacional de Ecologia, Universidad Nacional adjacent to Braulio Carrillo National Park, Costa Rica. Human Ecol-
Autonoma de Mexico, Mexico, D.F. ogy 34:407–431.
Ortiz, E., and J. Kellenberg. 2002. Program of payments for ecologi- Sierra, R., and E. Russman. 2006. On the efficiency of environmental ser-
cal services in Costa Rica. World Conservation Union, Washington, vices payments: a forest conservation assessment in the Osa Penin-
D.C. sula, Costa Rica. Ecological Economics 59:131–141.
Pagiola, S., J. Bishop, and N. Landell-Mills, editors. 2002. Selling forest Skole, D. L., and C. Tucker. 1993. Tropical deforestation and habitat frag-
environmental services: market-based mechanisms for conservation mentation in the Amazon: satellite data from 1978 to 1988. Science
and development. Earthscan Publications, London. 260:1905–1910.
Pagiola, S., K. von Ritter, and J. Bishop. 2004. Assessing the economic Van Laake, P., and G. A. Sánchez-Azofeifa. 2004. Focus on deforestation.
value of ecosystem conservation. Environment department papers, Agriculture, Ecosystems and Environment 102:3–15.
paper 101. The World Bank, Washington, D.C. Velkamp, E., A. M. Weitz, Starisky, and I.G. Huising. 1992. Deforesta-
Pfaff, A., and G. A. Sánchez-Azofeifa. 2004. Deforestation pressure and bi- tion trends in the Atlantic Zone of Costa Rica: a case study. Land
ological reserve planning: a conceptual approach and an illustrative Degradation and Rehabilitation 3:71–84.
application for Costa Rica. Resource and Energy Economics 26:237– Wilson K. A., M. F. McBride, M. Bodel, and H. P. Possingham. 2006.
254. Prioritizing global conservation efforts. Nature 440:337–340.
Sánchez-Azofeifa, G. A., R. C. Harriss, and D. Skole. 2001. Deforestation Zbinden S., and D.R. Lee. 2005. Paying for environmental services: an
in Costa Rica: a quantitative analysis using remote sensing imagery. analysis of participation in Costa Rica’s PSA program. World Devel-
Biotropica 33:378–384. opment 33:255–272.

Conservation Biology
Volume 21, No. 5, 2007