A Social Cost-Benefit Analysis of Conserving the Ranomafana- Andringitra-Pic d'Ivohibe Corridor in Madagascar

Abstract

We reviewed existing studies and data to conduct a social cost-benefit analysis for a new protected area in south-eastern Madagascar. T he global net present value of conserving the Ranomafana-Andringitra-Pic d'Ivohibe corridor is large and positive, with a mid range estimate of over US330million,ifbenefitsarenotweightedaccordingtotheincomeoftherecipient.However,atthelocallevel,netpresentvalueofconservationisnegativeataroundUS 330 million, if benefits are not weighted according to the income of the recipient. However, at the local level, net present value of conservation is negative at around US-1400 per household in forest frontier communities. A t the national level, net benefits are non-significant. Thus, for corridor conservation to make a positive contribution to global welfare these inequalities must be addressed. For this reason, and to ensure the future supply of biodiversity conservation, we recommend that measures to mitigate the costs of conserva tion be directed at all forest frontier communities and made contingent on the conservation of forests and biodiversity. Given the large global net benefit of conserving the corridor, we believe that these measures are affordable and sustainable.

Full text

A Social Cost-Benefit Analysis of
Conserving the Ranomafana-Andringitra-
Pic d’Ivohibe Corridor in Madagascar
By Neal J. Hockley & Ramy Razafindralambo
University of Wales, Bangor, UK & Conservation Intl., Madagascar
2006
30th December 2006
USAID Cooperative Agreement No. 687-A-00-04-00090-00
USAID strategic objective: SO6 - Biologically diverse forest ecosystems conserved
Project activity: Maintaining Biological Integrity of Critical Biodiversity Habitats (MIARO)
Prime recipient: Conservation International
Sub-recipients: WWF, WCS, ANGAP
Sponsoring USAID operating Unit: USAID/Madagascar, SO6
© Neal J Hockley & Ramy Razafindralambo 2006

i
A Social Cost-Benefit Analysis of Conserving the Ranomafana-
Andringitra-Pic d’Ivohibe Corridor in Madagascar
Neal J. Hockley1,2 and Ramy Razafindralambo3
1School of the Environment and Natural Resources, University of Wales, Bangor.
Deiniol Road, Bangor. LL57 2UW. Wales, United Kingdom.
neal.hockley@univ.bangor.ac.uk. Tel: +44 1248 382450 Fax: +44 1248 354997
2 Vokatry ny Ala, BP 1067, Fianarantsoa 301, Madagascar.
261 (0) 20 75 517 01
3 Conservation International, Antananarivo, Madagascar.
rrazafindralambo@conservation.org
Keywords: Andringitra; Benefits Transfer; Biodiversity; Conservation
and Development; Distribution of Benefits; Madagascar; Opportunity
Costs; Protected Areas; Ranomafana; Social Cost-benefit Analysis.
J. Economic Lit. Codes: Q20, Q23, Q24, Q25, Q26, Q28, Q34, Q54, Q57
Abstract
We reviewed existing studies and data to conduct a social
cost-benefit analysis for a new protected area in south-eastern
Madagascar. The global net present value of conserving the
Ranomafana-Andringitra-Pic d’Ivohibe corridor is large and
positive, with a mid range estimate of over US$ 330 million, if
benefits are not weighted according to the income of the
recipient. However, at the local level, net present value of
conservation is negative at around US$-1400 per household in
forest frontier communities. At the national level, net benefits
are non-significant. Thus, for corridor conservation to make a
positive contribution to global welfare these inequalities must
be addressed. For this reason, and to ensure the future supply
of biodiversity conservation, we recommend that measures to
mitigate the costs of conservation be directed at all forest
frontier communities and made contingent on the conservation
of forests and biodiversity. Given the large global net benefit of
conserving the corridor, we believe that these measures are
affordable and sustainable.
This report can be downloaded from: http://www.bangor.ac.uk/~afpe5d/SCBA.html

ii
Une Analyse Coûts-Bénéfices Sociale de la Protection du Corridor
Ranomafana-Andringitra-Pic d’Ivohibe, Madagascar
Neal J. Hockley1,2 et Ramy Razafindralambo3
1School of the Environment and Natural Resources, University of Wales, Bangor.
Deiniol Road, Bangor. LL57 2UW. Pays de Galles, Royaume Uni.
neal.hockley@univ.bangor.ac.uk. Tél. : +44 1248 382450 Fax : +44 1248 354997
2 Vokatry ny Ala, BP 1067, Fianarantsoa 301, Madagascar.
Tél. : +261 (0) 20 75 517 01
3 Conservation International, Antananarivo, Madagascar.
rrazafindralambo@conservation.org
Mots Clés : Andringitra ; transfert des bénéfices ; biodiversité ;
Conservation et développement ; répartition des bénéfices ;
Madagascar ; Coûts d’opportunités ; Aires protégées; Ranomafana;
Analyse coûts-bénéfices sociale.
J. Economic Lit. Codes: Q20, Q23, Q24, Q25, Q26, Q28, Q34, Q54, Q57
Résumé
Se basant sur l’étude des littératures et données existantes, nous
avons effectué une analyse coût-bénéfice sociale d’une nouvelle
aire protégée dans le sud est de Madagascar. Au niveau mondial,
la valeur actuelle nette de la conservation du Corridor
Ranomafana-Andringitra-Ivohibe est très importante et positive,
avec une valeur moyenne estimée à plus de US$ 330 millions, si
les bénéfices ne sont pas pondérés selon le revenu de la
bénéficiaire. Cependant, la valeur actuelle nette de la conservation
est négative, aux environs de US$-1400 par ménage, pour les
communautés locales aux abords de la forêt. Au niveau national,
le bénéfice net n’est pas significatif. Ainsi, pour espérer une
contribution positive de la conservation du corridor au bien être
global, ces déséquilibres doivent être considérées. Pour cette
raison et pour assurer la continuité de la conservation de la
biodiversité dans le future, nous recommandons que des mesures
pour alléger le coût de la conservation sont orientées vers toutes
les communautés aux abords de la forêt, en fonction de leur
performance sur la conservation de la forêt et de la biodiversité.
Etant donné l’importance du bénéfice net mondial de la
conservation du corridor, nous pensons que ces mesures sont
abordables et faisables sur long terme.
Ce rapport peut-être téléchargé au: http://www.bangor.ac.uk/~afpe5d/SCBA.html

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Extended Abstract
The Government of Madagascar plans to triple the size of the country’s protected
area network. Because of its rich biodiversity the forested corridor joining the
National Parks of Ranomafana and Andringitra and the Special Reserve of Pic
d’Ivohibe has been identified as a priority site for a new protected area.
We reviewed existing studies and unpublished data to carry out a social cost-benefit
analysis of conserving the entire corridor, including the existing two National Parks
and the Reserve. Based on the results of these studies, we evaluate the costs and
benefits of corridor conservation in terms of its impact on: local livelihoods; timber
production; irrigation; carbon sequestration and ecotourism. We also use an
innovative approach to value the international non-use values of biodiversity. Despite
considerable uncertainties at all levels, we find the global net present value of
conserving the corridor to be large and positive, with a mid range estimate of over
US$ 330 million (range: 22-642m), when benefits are not weighted according to the
income of the beneficiary.
We then determine the distribution of costs and benefits across different stakeholder
groups, from forest edge communities, through the national to the global level.
Despite being overwhelmingly positive at the global level, our review suggests that
the benefits of conservation are unequally distributed, with significant costs at the
local level. We estimate the mid-range net present value of costs to forest frontier
communities at over US$1400 (range: US$196-2610) per average household. These
costs will be distributed unequally among communities. At the national level, the
conservation of the corridor has a slightly positive net present value of US$13 million
(+/- 83m), insignificant compared to the size of the uncertainties.
These inequalities mean that the true contribution which corridor conservation makes
to global welfare is highly dependent on benefits being transferred to local
communities and to the Malagasy nation. Without such transfers, conserving the
corridor would most probably have a negative net effect on global welfare, and its
success would not be assured. Conservation has the potential to make a real positive
contribution to Madagascar’s development, but for this to happen, Madagascar must
capture a higher proportion of the international value of its forests.
Despite the very large quantitative uncertainties, the economic case for corridor
conservation is therefore compelling when the global, un-weighted, value is
considered. However, the case for transferring benefits to the local and national
levels is equally compelling: to ensure the success of conservation and to ensure it
makes a positive, rather than negative impact on human welfare. Ecotourism and
carbon sequestration credits are likely to play an important role in the long-term. In
the short-term, we advocate the use of direct and proportionate mitigation measures
that are compatible with, and contingent on the supply of, biodiversity conservation.
Because the market for biodiversity conservation is undeveloped at present, we
recommend continuing or increasing transfer payments for biodiversity, such as
donor financing and government support, whilst ensuring that they are well targeted
to those bearing the costs. Market-based mechanisms should also be developed.
Finally, we note that the local costs, while very significant in comparison with local
incomes, are small compared to the global benefits, and conclude that such
measures should be affordable.
iii

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Contents
Abstract i
Résumé (en français) ii
Extended Abstract iii
Contents iv
List of Figures and Tables v
Acknowledgements vi
Introduction 1
Policy context and objectives of the study 1
Social Cost-benefit Analysis 1
Methodology 4
Framework of the analysis 4
Methods 5
Spatial Coverage 5
Missing Values 8
Results 9
Global Value of Conserving the Corridor 9
Distribution of Net Benefits 9
Costs at the forest frontier 10
The size of the Strictly Protected Core Zone. 11
Southern Extension of the Corridor to Vondrozo 12
Logging in the Conservation Site 12
Uncertainty and sensitivity 13
Discussion 15
Costs and benefits, winners and losers 15
Assuring the supply of biodiversity conservation 15
Conclusions 17
References & Sources 18
Appendix 1: Full Methods I
Appendix 2: Full results table XXII
Appendix 3: Notes on hydrological functions of forests, by J. E.
Annis & N. J. Hockley XXIV
iv

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
List of Figures and Tables
Figures
Figure 1: Map of the Ranomafana – Andringitra – Pic d’Ivohibe Corridor 3
Figure 2: The effect of increasing the size of the Strictly Protected Core
Zone on NTFP opportunity costs
12
Figure A1. Spatial extent of NTFP collection IX
Tables
Table 1: Studies and data used in the analysis 6-7
Table 2: Missing values 8
Table 3: Global Net Present Values 9
Table 4: Distribution of costs by stakeholder group. 9
Table 5: Breakdown of costs and benefits at the fokontany level 11
Table 6: The biodiversity value of the Vondrozo extension 12
Table 7: The benefits of sustainable logging in 1/3 of the buffer zone. 13
Table 8: Lower, mid-range and upper net benefits by stakeholder level 13
Table 9: Sensitivity of mid-range global values to discount rate and time
horizon.
14
Table 10: Mechanisms for the capture and transfer of external benefits
of conservation
15
Table A1: Estimates of the pure rate of time preference II
Table A2: Estimates of per capita growth rates III
Table A3: Estimates of e the elasticity of marginal utility of consumption IV
Table A4: Estimates of social discount rates IV
Table A5: A comparison of per capita deforestation rates XI
Table A6. Timber production per hectare for two different methods. XVII
Table A7: Taxes on logging operations XVIII
Table A8 Labour costs in logging operations XVIII
Table A9: Breakdown of benefits from logging operations XIX
Table A10: Full breakdown of the results XXII
v

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Acknowledgements
This study was primarily financed by Conservation International through the MIARO
project of USAID financial assistance to Madagascar’s Third Environmental
Programme. Additional support was received from Ecoregional Initiatives,
Fianarantsoa.
The specialist groups at the workshop were led by: Mijasoa Andriamarovololona,
Joanna Durbin, Julia Jones, David Meyers, Julien Noel, Holy Raharinjanahary,
Vololoniaina Raharinomenjanahary, Anicet Ranaivoarison, Rija Ranaivoarison, Haja-
Guy Randrianarisoa, Florent Ravoavy and Georges Serpantié, we thank them for
their efforts.
We would also like to thank the following people for assistance with and
presentations at, the workshop: Vladimir Ratsimandresy, BAMEX, Fianarantsoa;
Rondro Andriamampionona, Michel Randriambololona and Jean Razafitsotra,
Conservation International, Fianarantsoa; Frank Hawkins and Fabi Randrianarisoa,
Conservation International, Antananarivo, Jean-Solo Ratsisompatrarivo, DAI,
Antananarivo, Mark Freudenberger of ERI, Fianarantsoa, Karen Freudenberger,
FCER, Fianarantsoa.
We thank Lisa Gaylord and the team at USAID for continuing technical and financial
support.
Thanks also to the following people or institutions who helped with requests for
documentation or information: ANGAP Ranomafana, Fianarantsoa and
Antananarivo; Jonathon Annis; Pieter van Beukering; Geoff Bright; JIRAMA
Fianarantsoa; David Knox; Susanne Menzel; Dominic Moran; Ny Tanintsika,
Fianarantsoa; Colin Price; Etienne Rasamisandy; SAGE, Fianarantsoa; Bert Saveyn;
Georges Serpantié and Richard Tol.
Finally, Frank Hawkins and David Meyers read an earlier draft of this report, and
made many helpful and insightful comments. The responsibility for any errors or
omissions which remain is of course our own.
Misaotra indrindra
vi

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Introduction
Policy Context and Objectives of the Study
At the World Parks Congress in Durban, SA, in 2003, President Marc Ravalomanana
announced Madagascar’s intention to triple the size of its protected areas. As a
result, the corridor of natural forest linking the National Parks of Ranomafana and
Andringitra and the Special Reserve of Pic d’Ivohibe (Figure 1, below) has been
identified as a key part of the expansion strategy1. The corridor is a hugely important
refuge for biodiversity and, together with the two parks, boasts two endemic primate
species2 and contains a wide range of rare and endemic flora and fauna. As a result,
the corridor will be designated a “Conservation Site3”, a new form of protected area.
While conservation organisations and the international community have welcomed
the President’s vision, they recognise the need to assess the socio-economic impact
of the new protected areas; to address the concerns raised by political leaders in the
corridor region; and to provide an up to date assessment of the economic value of
conserving the forest corridor.
Conservation International (CI) and Ecoregional Initiatives (ERI) responded to these
needs in two ways:
First, a workshop was held in Fianarantsoa, Madagascar, on the 26-28th April 2005,
to enable stakeholders to discuss the socio-economic implications of the new
Conservation Site, and to gather information and opinion on the development needs
of the surrounding communities. Around 80 delegates, including local community
leaders, researchers, representatives of NGOs and government agencies, attended
the workshop.
Second, we were asked to prepare a study that quantified the costs and benefits of
conserving the corridor, while also determining their distribution across local, national
and international scales. The study would build on the findings of the workshop, and
use existing studies and data rather than carry out new research.
This report presents the results of this study and a discussion of their implications.
Full details of the methodology can be found in Appendix 1. The original
spreadsheet, containing all of the calculations as well as most of the data used in the
analysis, is available from the authors. Anyone requiring further information is
welcome to contact the authors.
Social Cost-benefit Analysis
Social Cost-Benefit Analysis (SCBA) involves the quantification of the costs and
benefits of a project, with the aim that these can be aggregated to determine whether
the project makes a net positive or negative economic contribution to society. SCBA
should consider all costs and benefits, including those for which no markets exist. It
should also explicitly consider the distribution of costs and benefits amongst different
1 By a biological priority setting workshop held in Fianarantsoa in January 2005: L’Atelier Scientifique de
Planification: Corridor Forestier Ranomafana – Andringitra – Pic d’Ivohibe. 17-20 janvier 2005. Alliance
Ecorégionale (2005).
2 The Golden Bamboo Lemur (Hapalemur aureus) and the Greater Bamboo Lemur (Hapalemur simus).
3 Or “Site de conservation”. Since the study, the site has become “La Future Aire Protégée de Fandriana
– Vondrozo” which includes northern and southern extensions to Fandriana and Vondrozo respectively.
1

Introduction
stakeholders. In this manner, the equity implications of a project can be assessed,
along with its economic efficiency.
SCBA is only one of many tools available for appraising the desirability of projects,
including multi-criteria decision-making, rapid and participatory rural appraisal and
soft systems analysis (see Blumenthal & Jannink [2000] and Joubert et al. [1997] for
discussions of these methods). Each tool differs in the degree to which it achieves
the participation of stakeholders, the quantification of costs and benefits, and whether
it aims simply to inform the decision making process (e.g. cost-benefit analysis) or
whether it actually is a decision making process (e.g. participatory rural appraisal,
[Chambers 1994]).
It is fair to say that SCBA tends to be one of the less participatory decision making
tools. It is also true that SCBA has no explicit means of dealing with costs or benefits
that are extremely difficult to value in economic terms. Given this: why should we use
SCBA? We believe that SCBA is useful in this case, firstly, because many of the
most important costs and benefits have been quantified and secondly, because
despite advances in alternative methods, there exists no more satisfactory method
for aggregating diverse benefits, and the divergent preferences of many different
stakeholders (see for example Pearce [1998 p96-97]). We feel that the best solution
is to carry out an “open” SCBA, in which as much attention is directed to the
breakdown of costs and benefits, their distribution across groups and to those costs
and benefits which could not be evaluated economically, as is given to the overall
result. We feel that this represents the most useful way to present to decision makers
the information which is available.
2

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Figure 1: Map of the Ranomafana – Andringitra – Pic d’Ivohibe Corridor
showing existing protected areas and the southern extension.
3

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Methodology4
Framework of the Analysis
The cost-benefit analysis was designed to evaluate the economic case for
conserving the corridor. It aims to identify, and where possible to quantify, the
benefits and costs of protecting the forests of the corridor. It does this by comparing
two alternative scenarios:
• Scenario A: The null scenario, without conservation. Deforestation and extractive
activities continue as they would in the absence of protection.
• Scenario B: A Conservation Site is established, composed of a core of strictly
protected forest, surrounded by a zone in which certain extractive uses, but not
logging, are allowed.
In addition, the economic value of allowing sustainable, community-managed
selective logging within scenario B was assessed. However, it was impossible to
assess in a quantitative fashion many of the costs of allowing this exploitation. The
analysis is also designed to determine the distribution of these costs and benefits
across five key geographically determined stakeholder groups:
• Fokontany5 bordering the forest (Forest Frontier Fokontany [FFF]).
• Communes bordering the forest. (Forest Frontier Communes [FFC])
• Regions surrounding the corridor.
• Madagascar.
• The international community.
By assigning costs and benefits to the stakeholder groups to which they accrue, we
can identify winners and losers, and seek strategies that will improve the efficacy and
equitability of forest conservation.
One of the most significant costs which we evaluate at the local level, is the
opportunity cost of no longer being able to convert the corridor for agriculture (tavy6).
Much of this conversion has been illegal, since the colonial times, although its
illegality has often had little or no effect (see Kull 2004). However, residents of the
corridor maintain that they have rights to the corridor and that, while illegal, practising
tavy is one of those rights (Ferraro 1994, Kull 2004). We have conducted our
analysis with the assumption that residents of the corridor had the right, albeit only
customary, to practise tavy, and that policy makers wish to know what impact the
complete prohibition of tavy will have on their livelihoods.
4 Full details of the methodology can be found in Appendix 1. The original spreadsheets, containing all
of the calculations and data used in the analysis, together with most of the documents used in the study,
are available from the authors. In this section we summarise the main features of the SCBA, and the
information we used.
5 Fokontany are the lowest unit into which communities are organised by the government. Each
fokontany contains one or a few villages and fokontany populations usually range from 500-3000
inhabitants.
6 Some differentiate between tavy and tevy, with the latter reserved for the clearing of primary forest,
while the former refers to slash and burn of fallow vegetation. However, the term tavy has become
widely used, and any practitioners of tavy (or tevy) use tavy to mean clearing of forest.
4

Methodology
Methods
No resources were available for any primary studies of the costs and benefits of
conserving the corridor. We therefore proceeded by applying the results of previous
studies to the corridor. This approach, termed “benefits transfer” (Bateman et al.
2002) is of course open to criticism. Its accuracy will depend on the quality of the
original research, and the degree of similarity between the original situation, and that
prevailing in the area to which it is applied. However, where new, dedicated research
is impossible, benefits transfer represents a pragmatic compromise between
reliability and cost.
Table 1 (overleaf), shows the main values that we were able to quantify, and the
sources of information we used. In addition to the information mentioned in table 1,
we used basic information on the populations of communes and fokontany
surrounding the corridor, deforestation rates and remaining forest cover. In most
cases we have used two separate estimates, representing upper and lower estimates
of each benefit or cost. We have then used the mid range values in the results
presented below.
Spatial Coverage
The two national parks at either end of the corridor, Ranomafana and Andringitra,
together with the Special Reserve of Pic d’Ivohibe, are already protected and
managed by ANGAP, the National Parks service. The Durban Vision proposals relate
only to the corridor between the two parks and the reserve, and will not affect their
status. In economic terms, it would be preferable to determine the marginal value of
conserving additional forest outside of the parks, in order to determine the optimum
level of conservation. However, for both biological and political reasons, this is not
sensible. From a policy perspective, there are no proposals to conserve only a part of
the corridor7. Biologically, it is impossible to determine how biodiversity would be
affected by the loss of different areas of forest since the processes which determine
the resilience of biodiversity to habitat loss are poorly understood (du Toit et al.
2004). In any case, it would be even more difficult to determine how this would affect
the benefits that flow from biodiversity (see Tisdell et al. 2005 for an interesting
example, however). For these reasons, the analysis considers the Conservation Site
in the corridor and the parks together (Scenario B) and compares this to a situation
where no parks, reserves or Conservation Sites were present (Scenario A).
Although the biological priority-setting workshop recommended that the Conservation
Site should include the extension of the corridor as far as Vondrozo, we have
confined our analysis to the Ranomafana – Andringitra - Pic d’Ivohibe Corridor, for
reasons of data availability. However, our conclusions are likely to be qualitatively
true of this corridor extension (see Results and Discussion below).
7 Though different areas of the corridor may be protected to different degrees, with, for example,
selective logging being permitted in some areas but not others. We have tried to structure our analysis
in such a way that such plans could be evaluated.
5

Table 1: Studies and data used in the analysis
Value Information Used and Outline of the Calculations
Forest Products We used Ferraro’s (1994, 2002) estimates of the per-household opportunity costs of lost forest product
collection opportunities around Ranomafana National Park and applied them to the whole corridor. We then
used data on the spatial extent of a community’s forest product use (Vokatry ny Ala, unpublished data) to
determine the impact on local people’s forest product collection of different sizes of strictly protected core
zones (Scenario B), and of losing forest to deforestation in Scenario A.
Tavy We used Ferraro’s (1994, 2002) estimates of the per-household opportunity costs of lost tavy opportunities
around Ranomafana National Park and applied them to the whole corridor. We also used the estimates of
Kremen et al. (2000) of per hectare benefits of tavy, and applied them to the corridor using corridor specific
deforestation rates.
Ecotourism Data on ecotourism revenues (ANGAP [unpublished data], VOI ANJA [unpublished data]) are used together
with forecasts of ecotourism growth (World Tourism Organisation, quoted in Carret & Loyer [2003]), to
estimate the value of ecotourism in the corridor, into the future. Data on the distribution and magnitude of
indirect benefits are also used (Carret and Loyer [2003], ANGAP [No Date].)
Irrigation The willingness to pay of rice farmers in another area of Madagascar to prevent upstream deforestation was
determined by Brand et al. (2002). We applied this to the agricultural populations of forest edge communes,
using population data from the ILO census (Minten et al. 2003).
Timber Data provided by workshop participants, and other informants, were used to estimate the benefits from two
different timber harvesting regimes in the corridor. The first was a multiple rotation (30yr) approach, in which
only good quality timber of adequate diameter was extracted. The second was a one-time approach in which
all economically useful timber was extracted in a single rotation.

Carbon
sequestration
van Kooten et al. (2004) reviewed carbon offset programs and we use their mean estimate of the amount of
CO2 released when tropical forest is cleared as our upper estimate. Our lower estimate is that provided by
Razafindralambo (unpublished data) for the Zahamena-Mantadia corridor. We combine these estimates with
estimates of deforestation rates (Miaro, [2005]), to determine the amount of CO2 release which is avoided
each year by conserving the corridor. In order to put an economic value on these avoided emissions, we use
two different estimates of the value of avoided CO2 emissions. The higher estimate is the mean full social cost
of CO2 emissions estimated by Tol (2005). The lower estimate is the median, again estimated by Tol (2005).
Bio-prospecting
values
For the value of the corridor that might be realised through bio-prospecting for new pharmaceuticals, we use
the value adopted, after review of the literature, by van Beukering et al. (2003) for tropical primary forest in
Indonesia.
Non-use values
of biodiversity
We use two estimates of the willingness to pay (WTP) of rich world citizens to conserve biodiversity or tropical
rainforests, which represent (largely) non-use values. Kramer & Mercer (1997) estimated the WTP of
American citizens in the early ‘90s to preserve (in addition to existing reserves) a further 5% of tropical
rainforests. This value was extrapolated to the rest of the rich world’s population (OECD [2004, p200]), and
the corridor’s “share” estimated on a per-area basis, using data on world tropical forest cover from FAO
(2001).
A second, higher, estimate came from Menzel (2005) who estimated the WTP of German households for the
preservation of critically endangered species. Once again this value was extrapolated for the rest of the rich
world’s population, and the corridor’s share calculated using the proportion of the world’s critically endangered
mammals and birds which are endemic to the corridor (Randrianasolo [pers com], IUCN [2004] quoted in
Mittermeier et al. [2004]).

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Missing Values
We wish to be as honest as possible about the limitations of this study and to this
end, we have tried to identify all major costs and benefits which we have failed to
quantify. These are listed in Table 2, and discussed below.
Table 2: Missing values, affected stakeholders, and their likely effect on
the value of the Conservation Site.
Missing Value Stakeholders
Affected
Likely Effect on Value
of Conservation Site
Through Travel1 FFF, FFC -/=2
Mining FFF, FFC, Regions,
Nation
---
Cultural & non-use values3 of
Malagasy citizens
FFF, FFC +
Drinking Water FFF, FFC +/=4
Other Hydrological Values Regions +/=5
1 The corridor is criss-crossed by paths used by communities on both sides for transporting
goods to and from market.
2 If through travel was prohibited by the Conservation Site, this would represent a cost to local
communities, but otherwise there would be no effect.
3 These are discussed in detail in Appendix 1.
4 There may be some effect of deforestation on drinking water availability, but this is likely to
be small and limited to villages on the forest frontier (J. Annis pers. com., Bruijnzeel 1990,
2004).
5 We may have failed to capture some hydrological values of the corridor, however, these are
likely to be small and / or dependent on certain key places remaining forested, rather than the
whole corridor. It is therefore difficult to estimate the difference between the two scenarios
(see discussion above, Bruijnzeel 1990, 2004 and Chomitz & Kumari 1996, 1998 and
Appendix 3, below).
In summary, the prohibition of through travel could add considerably to the local
costs of the Conservation Site, but would be unlikely to alter the qualitative results of
the analysis, and is, in any case, not a prerequisite for the establishment of the
Conservation Site. We would expect the other hydrological values to be of minor
significance (Chomitz & Kumari 1998, see also Appendix 3 below). This leaves:
cultural or non-use values; and mining. Cultural values may decrease local costs, but
are unlikely to reverse the qualitative nature of our results, for reasons discussed in
Appendix 1 p36). Many are, however, compatible with conservation, and there are no
reasons why there should be any conflict between the Conservation Site and most
traditional and cultural values of residents, provided that the site is designed
sensitively and in participation with local people (Appendix 1 p36).
Mining however is a huge unknown. The foregone benefits of mining depend on the
unknown reserves which are present beneath the corridor, and can only be estimated
with the aid of detailed geological studies. We believe it is most honest to exclude
mining from the analysis at this stage, but feel that the analysis would be useful to
decision makers considering an application for mining in the corridor, since it
evaluates most other benefits. The distribution of benefits from mining projects is also
hugely uncertain. Experience shows that all too often local communities benefit little
from mining in their area, and may bear significant costs in the form of land
degradation, water pollution and insecurity.
8

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Results
Below we describe the key results of our analysis. Unless otherwise stated, all
figures are mid-range values and assume that the Strictly Protected Core Zone of the
Conservation Site is delimited at 3km from the forest edge, and that no logging of any
form takes place in the corridor. Net Present Values are calculated over 60 years,
with a social discount rate of 5%8. The effect of changing these assumptions and
parameters is discussed below.
The full results table, giving complete breakdown of costs and benefits by type, and
stakeholder group, is given in Appendix 2.
Global Value of Conserving the Corridor
We estimate the global net present value of conserving the Ranomafana –
Andringitra - Pic d’Ivohibe corridor to be just over US$ 330 million over 60 years,
using a discount rate of 5%. In other words, conserving the corridor, along with the
three existing protected areas, represents a net benefit to the world, of US$ 330m,
compared with the null scenario of continued exploitation for slash and burn
agriculture and logging.
The upper and lower limits of this global NPV are US$ 22m and 642m respectively
(Table 3, below), demonstrating that even with the most conservative assumptions,
conservation of the corridor outperforms the null scenario by a substantial margin.
Table 3: Global Net Present Values (million US$)
Lower estimate
Mid-range
Estimate Upper Estimate
Global Net
Present Value 22 332 642
Distribution of Net Benefits
However, we are also interested in the distribution of costs and benefits among
different stakeholder groups: who are the winners and who are the losers? Table 4,
below, shows the distribution of benefits between each stakeholder group.
Table 4: Distribution of costs by stakeholder group.
Stakeholder groups
Cumulative
Net Benefit
mUS$
Net Benefit (US$) per
head of population
Forest Frontier Fokontany (FFF) -29 -168.49
Total Forest Frontier Communes (FFC) inc FFF -47 -94.82
Total National Benefit (inc FFF and FFC) 13 0.76
Total Global Benefit 332
Each row shows the cumulative net benefits, at that level. For example, the Net
Benefit at the National level sums the local costs, and national benefits, to give a
small positive National Net Benefit.
8 This rate is in line with Social Discount Rates calculated for India (5.2%, Kula 2004) and used in
Indonesia (4%, van Beukering et al. 2003). See Appendix 1, pp21-26 for justification and calculations.
9

Results
At the level of the Forest Frontier Fokontany and Communes, costs exceed benefits,
and costs are highest, per capita, in communities closest to the forest. At the national
level, mid-range benefits marginally exceed costs: while rural corridor populations
bear most of the costs of conservation, most of the benefits to Madagascar occur at
larger spatial scales.
These costs and benefits are of course averaged over each stakeholder group, and
mask the very large inter-household variance which must surely exist.
In addition these figures do not take account of the large quantities of development
assistance that the corridor receives (though they do take account of direct spending
on existing and proposed protected area management). In some areas, agricultural
assistance has greatly increased household incomes, while larger scale projects
such as the rehabilitation of transport networks may have boosted the regional
economy. However, we are aware of no studies which have evaluated the benefit to
rural communities of receiving development assistance. It is therefore impossible to
use the amount of funds spent on international development to estimate the net
benefit, in terms of income growth, for stakeholder groups.
In addition, like the costs and benefits of forest protection themselves, the benefits of
development assistance are not evenly distributed. It is likely that remote
communities, who have often benefited least from development initiatives to date,
also stand to gain least from the protection of the forest – they are unlikely to attract
ecotourists, and may find it harder to move away from tavy, into, for example cash
crops (see Nambena 2003).
The marginal utility of income, and the importance of compensation
In calculating the global value of conserving the corridor, we have assumed constant
marginal utility of income, i.e. that a dollar is worth the same to a poor Malagasy as it
is to a rich Westerner. Under these assumptions, conservation of the corridor would
increase aggregate global welfare (by 300m dollars or so). However, economists
recognise that the value of a unit of income is not the same for all people. This effect,
known as the “diminishing marginal utility of income”, means that income changes
which affect poor people should be weighted more highly than those which affect the
rich. In the case of the corridor, where costs are borne by the poor, and benefits
accrue to the rich, this effect will substantially reduce the net value of conservation. In
fact, using fairly conservative assumptions, it is easy to show that the net value of
conservation would be strongly negative9, if benefits are not transferred from (rich)
winners to (poor) losers or in economic parlance, if “compensation” is not paid. It
cannot be stressed too strongly that, evaluated in terms of welfare, the positive net
value of conservation is highly dependent on benefits being transferred and would
almost certainly be negative were this not to be the case. The figure of 330m global
NPV therefore assumes that winners compensate the losers.
Costs at the Forest Frontier
The most important net costs are felt at the level of those Fokontany which border
the forest. Table 5, below, shows the breakdown of costs and benefits by type, and
the average values per household. The opportunity costs of stopping tavy and timber
exploitation represent the biggest costs, while ecotourism provides the largest
benefits. With a buffer zone set at 3km from the forest edge, the opportunity costs of
9 For example, using a conservative value (1) for the elasticity of marginal utility of income, and
weighting costs and benefits according to the relative incomes of the stakeholder groups they affect,
turns the positive global net present value of US$330m negative by at least the same order of
magnitude.
10

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
lost forest product collection are small (see next section). Annualised costs per
household are significant, given average household incomes in the region.
Table 5: Breakdown of costs and benefits at the fokontany level
For the breakdown of costs and benefits at other levels, see Appendix 2 pp 42-43.
Benefit or Cost Net Present Value
of Benefit (US$)
Net Present
Value per
Average
Household
(US$)
Annualised Net
Benefit per
Average
Household (US$)
Non-Timber Forest products -2,288,436 -108.97 -5.76
Tavy -10,918,526 -519.94 -27.47
Ecotourism 13,148,280 626.12 33.08
Irrigation 860,320 40.97 2.16
Timber -30,268,587 -1,441.38 -76.15
Total -29,466,950 -1,403.21 -74.13
Note that if timber harvesting is allowed in the conservation site, even at low levels,
and if the local communities could control revenues (whereas traditionally they have
only been used as local labour) then the average benefits may be more likely to
approach average costs at the local level. Again it is important to note that these are
average costs and benefits, and the benefits from ecotourism, in particular, are likely
to be very patchily distributed.
The Size of the Strictly Protected Core Zone.
The Government of Madagascar has not yet decided the precise form that
Conservation Sites will take. However, it is likely that they will follow the model of
many protected areas by having two zones: a central strictly protected core
surrounded by an outer “buffer zone” (Razafitsotra pers. com.). In the central core,
only a few activities may be permitted, such as scientific research and access to
tombs, while in the buffer zone, certain sustainable extractive activities, such as non-
timber forest product (NTFP) collection, may be permitted. The relative size of these
two zones will be important in determining the impact of the Conservation Site on
local populations, as well the benefits in terms of conservation outcomes.
Using information on the distance travelled by forest product harvesters in the
Ranomafana area, we evaluated the effect of having core zones of different sizes on
the opportunity costs (due to lost NTFP collection opportunities). Figure 2 below,
summarises the effects on NTFP opportunity costs of decreasing the size of the
buffer zone, from 5km from the forest edge, to just 1km, and finally having no buffer
zone at all (100% strictly protected area).
As the size of the buffer zone increases, the costs of conservation in terms of lost
NTFP harvests increases. There is a large jump in opportunity costs between two
and one kilometres from the forest edge.10
10 Even with very large buffer zones, the benefits for NTFP collection of stopping deforestation do not
outweigh the costs of the strictly protected core zone, over the corridor as a whole, including the national
parks, where little natural forest remains outside of the strictly protected area. If the national parks are
excluded, there is a small benefit, in terms of NTFPs, for corridor households, if the buffer zone is set at
5km. However, these households still suffer net costs overall, because of other opportunity costs (e.g.
stopping tavy).
11

Results
Figure 2: The effect of increasing the size of the Strictly Protected Core
Zone, on NTFP opportunity costs
The Costs of Different Noyau Dur Regimes
0
5
10
15
20
25
30
35
40
5 4 3 2 1 No Forest Product
Collection Permitted
Distance of Noyau Dur from Forest Edge
Annualised Net Cost per Forest Edge Household
(US$)
Southern Extension of the Corridor to Vondrozo
The biological priority-setting workshop identified the southern extension of the
corridor, as far as the first major break at Vondrozo, to be of conservation
importance. This southern extension includes large areas of lowland (<800m) forest,
a habitat type very rare in the rest of the corridor. It also covers most of the global
range of a critically threatened lemur, E. albocollaris. However, because of the lack of
information about this area, we have restricted our analysis to the Ranomafana-
Andringitra-Ivohibe corridor and we cannot evaluate many of the costs and benefits
of this southern extension. However, it is possible to determine some of the
international benefits of extending the Conservation Site southwards. Table 6, below,
shows the added value, in terms of the non-use values of biodiversity, of conserving
the Vondrozo extension. Extending the conservation site would increase its
biodiversity value, in economic terms, by up to 50%, to US$ 375m.
Table 6: The biodiversity value of the Vondrozo extension
Ranomafana-Andringitra-
Ivohibe Corridor
Vondrozo
Extension Combined
International Non-Use
Values of Biodiversity
(Mid Range Values,
US$m)
237 138 375
Logging in the Conservation Site
The results presented above assume that no logging will be permitted in the corridor
under Scenario B. Using estimates provided by experts (workshop participants,
Rasamisandy pers. com.) we estimated the economic benefits of permitting low
impact, sustainable logging, in some parts of the corridor. For this we used the same
12

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
productivity per hectare per year as in the multiple rotation approach considered for
scenario A (see above, and Appendix 1). However, we assumed a rotation length of
40, rather than 30 years, and further assumed that logging was only permitted in 1/3
of the buffer zone (0-3km from the forest edge). As for Scenario A, no machinery is
used, and all timber is felled and extracted by hand. Table 7, below, shows our
estimate of the benefits of such logging, and of the potential reduction in opportunity
costs which such logging would produce.
Table 7: The benefits of sustainable logging in 1/3 of the buffer zone.
Stakeholder Group
Cumulative
NPV of
Conservation
Site (without
logging)
NPV of
Logging
Benefits in
1/3 of buffer
zone
NPV of
logging
benefits
per Ha
maximum
potential %
reduction
in
opportunity
costs
Forest Frontier Fokontany -29,466,950 2,784,423 40 9%
Forest Frontier Communes -46,546,387 5,662,283 82 12%
Corridor Regions -40,606,009 6,507,888 94 16%
National 12,782,933 8,783,079 127
Total Net Benefit World 332,101,942 8,783,079 127
All values are US$
All values are cumulative, i.e. the value for Forest Frontier Communes, includes all of the Commune
population, including those in the Forest Frontier Fokontany.
It cannot be stressed too strongly, that these are only the net benefits associated
directly with logging and we are unable to make any assessment of the effect of
allowing such logging on other values. For example, logging would affect biodiversity
(Ganzhorn et al. 1990); carbon sequestration (Healey et al. 2000) and possibly
ecotourism, although it would be unlikely to have any significant effect on
hydrological function of the forest (Bruijnzeel 1990, 2004, Serpantié pers. com.). It is
beyond the scope of this study to assess the likely magnitude of such effects. It is for
this reason that the final column of the table is titled: potential reduction in opportunity
costs. Permitting such logging may well reduce the global value of the corridor, as
well as reducing the ability of local communities to benefit from ecotourism.
Therefore, while permitting logging represents one way in which local opportunity
costs could be reduced, particularly in areas with little ecotourism potential, it is by no
means certain that it represents the most efficient solution.
Uncertainty and Sensitivity
Uncertainty
As with any similar study, considerable uncertainty exists as to the magnitude of net
benefits at each stake holder level. Table 8 below shows the lower, upper and mid-
range estimates of Net benefits at key stakeholder levels.
Table 8: Lower, mid-range and upper net benefits by stakeholder level
Stakeholder Level
Lower
Estimate US$
Mid-range
Estimate US$
Upper Estimate
US$
Total Benefit to Forest
Frontier Fokontany -54,808,473 -29,466,950 -4,125,426
Total Net Benefit to Forest
Frontier Communes -87,393,271 -46,546,387 -5,699,502
Total National -70,785,475 12,782,933 96,351,341
Total Net Benefit World 22,078,510 332,101,942 642,125,374
All values are cumulative, i.e. the value for Communes includes the Forest Frontier Fokontany
13

Results
This table shows that while the magnitude of net benefits is highly uncertain at all
levels, the sign is robust at the local and global level. The positive net benefit at the
national level is, however, marginal. Our mid-range estimate is positive, but this is
highly vulnerable to changes in parameter values and in discount rate and time
horizon (discussed below). Other studies have shown negative net present values at
the national level for conservation elsewhere in Madagascar, (Kremen et al. 2000 for
Masoala). It should be stressed again that no primary research was possible for this
study, and therefore, the robustness of its conclusions depend on the quality of the
original research and expert opinions, as well as the reliability of transferring values
from one study to another. Considerable uncertainty therefore exists over the
quantitative level of net benefits at all scales; however, the qualitative findings at the
local and global level appear robust.
Sensitivity to Discount rate and Time horizon
All the results above are calculated with a 5% discount rate and 60 year time horizon.
Table 9 below shows the effect on the mid-range global value of reducing the time
horizon to 40 years, and using a range of discount rates. As is commonly the case,
lengthening the time horizon, and / or decreasing the discount rate tends to favour
the case for conservation, but the global value is actually rather robust to changes in
these key parameters. One reason for this is that both estimates of biodiversity value
are one-off statements of value, essentially Net Present Values, and enter into the
calculations only in year 1.
Table 9: Sensitivity of mid-range global values to discount rate and time
horizon.
Time Horizon
Discount Rate 40 60
0.03 330,151,776 448,267,445
0.05 287,432,396 332,101,942
0.08 257,073,241 268,067,452
0.1 247,515,481 251,991,010
At the national level, The NPV of Scenario B turns negative at discount rates of
greater than 5.5%, (60yr) and 2% (40 yr) – showing how tenuous this positive
national value is. This is because Madagascar captures only a small percentage of
the global benefits of its conservation programs (see discussion). The NPV of
scenario B at the fokontany level is negative at all discount rates using either a 40 or
60 year time horizon. Overall then, the conclusion of global benefits and local costs
are extremely robust to discount rates, while the net benefit at the national level is
tenuous.
14

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Discussion
Costs and Benefits, Winners and Losers
We have shown substantial global benefits of conservation, accompanied by
significant local costs. In this, our results are unexceptional. The same story has
been told by other studies here in Madagascar11 (Ferraro 2002, Kremen et al. 200012,
Minten 2003, Shyamsundar & Kramer 1997), and elsewhere (see Balmford & Whitten
2003 for a review).
The ubiquity of the local costs of conservation is one reason for biodiversity loss
worldwide. The global benefits of conservation, on the other hand, provide the
justification for its conservation and for support from the international community.
Global net benefits mean that we can and should conserve the corridor. Local
costs mean that the corridor will not preserve itself, and that we must take
deliberate action. The supply of biodiversity conservation, like that of any other
good or service, will only be assured if its benefits are captured where they are
highest (at the international level) and its costs are paid where they too are highest
(at the local level).
Assuring the Supply of Biodiversity Conservation
Having identified the problem, which is the best mechanism to capture external
benefits and transfer them to the local level, thus ensuring the supply of biodiversity
conservation in the corridor? There are many ways in which we can help local
communities to conserve the corridor and to compensate them for the costs of doing
so. Some mechanisms offer a way of capturing the external values of biodiversity,
others a way of transferring that value to local communities (see Table 10 below),
and a few do both. A mixture of mechanisms will almost always be needed.
Table 10: Mechanisms for the capture and transfer of external benefits of
conservation
Mechanism Captures Benefits Transfers Benefits
Ecotourism 3 3
Carbon Sequestration 3 2
Global Environment Facility 3 3
Development Assistance 2 3
Community Based
Conservation
3? 3
Direct Incentives 2 3
Ecotourism has proved successful at Anja, on the western side of the corridor and
contributes to the costs of conserving Madagascar’s existing protected areas (Peters
1998, ANGAP no date). However, our analysis shows that even with optimistic
11 Some of these studies have been used in this study – therefore, our conclusions are not entirely
independent from theirs.
12 Although Kremen et al. (2000) show net benefits at the local level, this is only once development
actions aimed at compensating villagers are taken into account.
15

Discussion
predictions of growth in corridor ecotourism, it will be many years before it can play a
substantial part in conserving the whole corridor.
Through the Kyoto protocol, developing countries can internalise some of the global
benefits of carbon sequestration in their forests (Niesten et al. 2002). However,
Madagascar is unlikely to realise more than a fraction of the true value of conserving
the corridor through this mechanism. Furthermore, the bureaucratic hurdles involved
mean that it may be several years before any money flows into the corridor from
carbon sequestration. The Global Environmental Facility (GEF), administered by
the World Bank and the UNDP, provides another mechanism by which rich countries
can assist poor countries to protect their biodiversity. In doing so, it captures some of
the substantial benefits which rich country citizens currently obtain free of charge
(Menzel 2005). The projects it funds, in biodiversity rich countries, transfer these
benefits to those bearing the costs of conservation.
Development assistance can significantly raise the incomes of forest edge
communities, as well as making them less dependent on activities such as tavy,
which conflict with conservation (LDI 2004). To be successful in achieving
conservation, these benefits need to be made contingent on conserving the forest,
and the links between development assistance and conservation have not always
been tight enough to ensure this (Ferraro 2001). They may nevertheless be crucial to
creating the wider conditions necessary for conservation on a regional level
(Freudenberger 2003).
Community Based Conservation can help to increase the local benefits of
improved natural resource management, but perhaps more importantly it can provide
a vehicle for capturing and transferring the benefits of conservation at all scales, to
those communities directly responsible for conserving the forest. This could be
through enabling ecotourism, carbon sequestration, or providing a mechanism for the
payment of direct incentives for biodiversity conservation (below). In this case,
communities would be paid for their biodiversity conservation, which may go well
beyond that demanded by their members, and even as far as strict protection.
A final mechanism is direct incentives for conservation (Durbin et al. 2001, Ferraro
& Kiss 2002). Communities are paid, in cash or in kind, for achieving biodiversity
conservation goals. Funding comes from conservation NGOs, the GEF, or carbon
sequestration. This approach is relatively new, but has been trialled in the Menabe
region of Madagascar (Durrell Wildlife Conservation Trust, 2005). The payment of
direct incentives is likely to be complex where communal forests are concerned, and
may rely on the presence of strong community management, as in the Durrell case.
With any mechanism, sustainability is a concern. However, it is important not to
confuse sustainability with self-sufficiency. The world derives enormous external
benefits from biodiversity conservation in Madagascar, and these are not likely to
decrease in the near future. Given this, and the poverty of the country, there is no
reason why conservation should be locally self-sufficient, or self-financing, for many
years to come. To restrict our search to approaches which require no outside
funding, or require it for only a limited period, is to make our task unnecessarily
difficult.
16

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Conclusion
We have demonstrated a robust economic argument for the conservation of the
Ranomafana – Andringitra – Pic d’Ivohibe corridor. Globally, society stands to gain
by tens of millions of dollars annually as a result of the protection of the
corridor.
However, the benefits of forest protection are divided extremely unequally,
with the heaviest costs being borne by some of the world’s poorest people.
Although these costs can be minimised by carefully designing the conservation site in
participation with local people, substantial costs will remain. If these are not
compensated for, global welfare may suffer as a result of conservation.
Historically, no mechanism has existed to transfer benefits from those who gain from
the forest’s existence, to those who threaten it (Balmford & Whitten 2003). This has
led to a drastic undersupply of forest conservation. Today, although many are
relatively untested, these mechanisms exist, and the future of the corridor, as well as
the livelihoods of the people who live around it, will depend on them. Whichever
methods are chosen, they must reach all areas of the corridor, and all sectors of the
population who are affected. Furthermore, any benefits must be contingent upon the
sound management and long lasting conservation of the corridor.
The existence of a positive global net present value for the preservation of the
corridor represents both the argument for conserving it, and the means to do
so. The corridor can be conserved - but only if we succeed in capturing the
enormous external benefit of the corridor’s conservation and transferring it to
local communities, and the nation of Madagascar.
17

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
References and Sources
Quoted in the main report and appendices. Electronic copies of most documents are
available on request from the 1st author.
Personal communications
Annis, J. ERI, Fianarantsoa, Madagascar. August 2005.
Rasamisandy, E. CIREEF, Eaux et Forêts, Fianarantsoa, Madagascar. August 2005
Serpantié, G. IRD, Antananarivo, Madagascar. April 2005
Randrianasolo, H. Conservation International, Antananarivo, Madagascar
Razafitsotra, J. Conservation International, Fianarantsoa, Madagascar. April 2005.
Unpublished data
ANGAP [unpublished data]. Available from the first author.
Razafindralambo [unpublished data]. Available from the second author.
VOI ANJA [unpublished data]. Available from the first author.
Vokatry ny Ala [unpublished data]. Available from the first author.
References
Alliance Ecorégionale (2005). Acte de l'Atelier Scientifique de Planification: Corridor
Forestier Ranomafana - Andringitra - Pic d'Ivohibe pp 27.
ANGAP (No Date). Evaluation de l'Impact Economique des Aires Protégées
Balmford, A. & Whitten, T. (2003) Who should pay for tropical conservation, and
how could the costs be met. Oryx, 37, 238-250.
Bateman, I.J., Carson, R.T., Day, B., Hanemann, W.M., Hanley, N., Hett, T.,
Jones-Lee, M., Loomes, G., Mourato, S., Özdemiroglu, E., Pearce, D.W.,
Sugden, R., & Swanson, J. (2002) Economic Valuation with Stated Preferences
Techniques: A Manual. Edward Elgar, Cheltenham, UK.
Blamey, R. (1996). Citizens, consumers and contingent valuation: clarification and
the expression of citizen values and issue-opinions. In Forestry, Economics and the
Environment (eds W.L. Adamowicz, P.C. Boxall, M.K. Luckert, W.E. Phillips & W.A.
White). CAB International, Wallingford.
Blue, E.N. & Tweeten, L. (1997) The estimation of marginal utility of income for
application to agricultural policy analysis. Agricultural Economics, 16, 155-169.
Blumenthal, D. & Jannink, J.L. (2000) A classification of collaborative management
methods. Conservation Ecology, 4, art. no.-13.
Brand, J., Healy, T., Keck, A., Minten, B., & Randrianarisoa, C. (2002). Truths and
myths in watershed management: the effect of upland deforestation on rice
productivity in the lowlands, Policy Brief. FOFIFA Programme ILO. August 2002. pp
5.
Brand, J., Healy, T., Keck, A., Minten, B., & Randrianarisoa, C. (2002). Réalités et
mythes sur l'aménagement des bassins versants : l'effet de la déforestation des
pentes sur la productivité de riz dans les plaines, Policy Brief. FOFIFA Programme
ILO. August 2002. pp 5.
Bruijnzeel, L.A. (1990). Hydrology of moist tropical forests and effects of conversion:
a state of knowledge review. UNESCO, Paris. pp 225.
Bruijnzeel, L.A. (2004) Hydrological functions of tropical forests: not seeing the soil
for the trees? Agriculture, Ecosystems and Environment, 104, 185–228.
18

References & Sources
Carret, J.-C. & Loyer, D. (2003). Madagascar protected area network sustainable
financing economic analysis perspective, Paper Presented at the World Parks
Conference, Durban, South Africa, 11th-15th September, 2003. pp 11.
Chambers, R. (1994) The Origins and Practice of Participatory Rural Appraisal.
World Development, 22, 953-969.
Chomitz, K.M. & Kumari, K. (1996). The Domestic Benefits of Tropical Forests A
Critical Review Emphasizing Hydrological Functions, POLICY RESEARCH
WORKING PAPER Rep. No. 1601. World Bank. May 1996.
Chomitz, K.M. & Kumari, K. (1998) The domestic benefits of tropical forests: A
critical review. World Bank Research Observer, 13, 13-35.
Dixon, J.A., Carpenter, R.A., Fallon, L.A., Sherman, P.B., & Manopimoke, S.
(1986) Economic analysis of the environmental impacts of development projects
Earthscan Publications, London.
du Toit, J.T., Walker, B.H., & Campbell, B.M. (2004) Conserving tropical
nature:current challenges for ecologists. Trends in Ecology & Evolution, 19, 12-17.
Durbin, J., Andrianarimisa, A., & DeCosse, P.J. (2001). Le Potentiel des Contrats
de Conservation pour Contribuer à la Conservation de la Biodiversité à Madagascar -
The Potential of Conservation Contracts to Contribute to Biodiversity Conservation in
Madagascar, Unpublished Report to PAGE, Madagascar November 2001. pp 36.
Durrel Wildlife Conservation Trust (2005). Suivi écologique participatif 2004 de la
forêt du nord de Menabe central. Durrell WCT., Antananarivo, Madagascar.
ESRI (2002). ArcView GIS. Vers: 3.3.
Evans, D.J. (2005) The elasticity of marginal utility of consumption: Estimates for 20
OECD Countries. Fiscal Studies, 26, 197-224.
FAO (2001). Global Forest Resources Assessment 2000 Main report. Rep. No. 140.
FAO, Rome.
Ferraro, P.J. (1994) Natural Resource Use in the Southeastern Rain Forests of
Madagascar and the Local Impacts of Establishing the Ranomafana National Park.
Masters, Duke University.
Ferraro, P.J. (2001) Global Habitat Protection: Limitations of Development
Interventions and a Role for Conservation Performance Payments. Conservation
Biology, 15, 990-1000.
Ferraro, P.J. (2002) The local costs of establishing protected areas in low-income
nations: Ranomafana National Park, Madagascar. Ecological Economics, 43, 261-
275.
Ferraro, P.J. & Kiss, A. (2002) Direct payments to conserve biodiversity. Science,
298, 1718-1719.
Freudenberger, K.S. (2003). The Fianarantsoa-East Coast railroad and its role in
Eastern forest conservation. In The natural history of Madagascar (eds S.M.
Goodman & J.P. Benstead), pp. xxi+1709. University of Chicago Press, Chicago &
London.
Ganzhorn, J.U., Ganzhorn, A.W., Abraham, J.P., Andriamanarivo, L., &
Ramananjatovo, A. (1990) The Impact of Selective Logging on Forest Structure and
Tenrec Populations in Western Madagascar. Oecologia, 84, 126-133.
Green, G.M. & Sussman, R.W. (1990) Deforestation History of the Eastern Rain
Forests of Madagascar from Satellite Images. Science, 248, 212-215.
Hadker, N., Sharma, S., David, A., & Muraleedharan, T.R. (1997) Willingness-to-
pay for Borivli National Park: Evidence from a contingent valuation. Ecological
Economics, 21, 105-122.
19

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Hawkins, F. & Horning, N. (2001). Forest cover change in USAID and control areas.
A preliminary evaluation and report to USAID and partners. PAGE, Antananarivo.
July 2001. pp 21.
Healey, J.R., Price, C., & Tay, J. (2000) The cost of carbon retention by reduced
impact logging. Forest Ecology and Management, 139, 237-255.
HM Treasury (2003) The Green Book - Appraisal and Evaluation in Central
Government HMSO, London.
Irwin, M.T., Johnson, S.E., & Wright, P.C. (2005) The state of lemur conservation
in south-eastern Madagascar: population and habitat assessments for diurnal and
cathemeral lemurs using surveys, satellite imagery and GIS. Oryx, 39, 204-218.
Joubert, A.R., Leiman, A., deKlerk, H.M., Katua, S., & Aggenbach, L.C. (1997)
Fynbos (fine bush) vegetation and the supply of water: a comparison of multi-criteria
decision analysis and cost-benefit analysis. Ecological Economics, 22, 123-140.
Kahneman, D. & Knetsch, J.L. (1992) Valuing public goods: the purchase of moral
satisfaction. Journal of Environmental Economics and Management, 22, 57-70.
Kramer, R.A. & Mercer, D.E. (1997) Valuing a global environmental good: US
residents' willingness to pay to protect tropical rain forests. Land Economics, 73, 196-
210.
Kremen, C., Niles, J.O., Dalton, M.G., Daily, G.C., Ehrlich, P.R., Fay, J.P.,
Grewal, D., & Guillery, R.P. (2000) Economic incentives for rain forest conservation
across scales. Science, 288, 1828-1832. See also Supplementary Information on CD
and Web.
Kula, E. (2004) Estimation of a social rate of interest for India. Journal of Agricultural
Economics, 55, 91-99.
Kull, C.A. (2004) Isle of fire. The political ecology of landscape burning in
Madagascar University of Chicago Press, Chicago & London.
Lambert, P.J., Millimet, D.L., & Slottje, D. (2003) Inequality aversion and the
natural rate of subjective inequality. Journal of Public Economics, 87, 1061-1090.
LDI (2004). Final Report March 2004.
Lean & Rowntree (1993) A GCM simulation of the impact of Amazonian
deforestation on climate using an improved canopy representation. Quarterly Journal
of the Royal Meteorological Society, 119, 509-530.
Menzel, S. (2005). Financial support for biodiversity protection in developing
countries – does the CBD mechanism lead to an appropriate level of biodiversity
protection? In Valuation and Conservation of Biodiversity Interdisciplinary
Perspectives on the Convention on Biological Diversity (eds M. Markussen, R. Buse,
H. Garrelts, M.A. Manez Costa, S. Menzel & R. Marggraf), pp. xxx, 430 p. 57 illus.
Springer.
MIARO (2005) Le corridor Ranomafana - Andringitra - Ivohibe et les principales
zones de deforestation. MIARO.
Minten, B. (2003) Compensation and cost of conservation payments for biodiversity.
Minten, B., Randrianarisoa, J.-C., & Randrianarison, L., eds. (2003) Agriculture,
pauvreté rurale et politiques économiques à Madagascar, pp 107. FOFIFA,
Antananarivo.
Mittermeier, R.A., Robles Gil, P., Hoffmann, M., Pilgrim, J., Brooks, T.,
Mittermeier, C.G., Lamoreux, J., & da Fonseca, G.A.B., eds. (2004) Hotspots
Revisited, pp 391. CEMEX.
Moseley, W. G. (2001) African Evidence on the Relation of Poverty, Time Preference
and the Environment. Ecological Economics, 38, 317-326.
20

References & Sources
Nambena, S.M. (2003) Using Geographic Information Systems to Assess
Possibilities of Expanding Traditional Agroforestry in Slash-and-Burn Zones in
Madagascar. Mountain Research and Development, 23, 329–333.
Niesten, E., Frumhoff, P.C., Manion, M., & Hardner, J.J. (2002) Designing a
carbon market that protects forests in developing countries. Philosophical
Transactions of the Royal Society of London Series a-Mathematical Physical and
Engineering Sciences, 360, 1-14.
OECD (2004) Quarterly Labour Force Statistics.
Pearce, D. (1998) Cost-benefit analysis and environmental policy. Oxford Review of
Economic Policy, 14, 84-100.
Peters, J. (1998) Sharing national park entrance fees: Forging new partnerships in
Madagascar. Society & Natural Resources, 11, 517-530.
Price, C. (1993) Time, Discounting and Value. Blackwell, Oxford.
Price, C. (No Date). Economic valuation of biodiversity: a consequentialist
pilgrimage, SAFS Monograph. University of Wales, Bangor, Bangor.
Sader, S.A. (1995) Spatial Characteristics of Forest Clearing and Vegetation
Regrowth as Detected by Landsat Thematic Mapper Imagery. Photogrammetric
Engineering and Remote Sensing, 61, 1145-1151.
Sharma, R.A., McGregor, M.J., & Blyth, J.F. (1991) The Social Discount Rate for
Land-Use Projects in India. Journal of Agricultural Economics, 42, 86-92.
Shyamsundar, P. & Kramer, R. (1997) Biodiversity Conservation - At What Cost? A
Study of Households in the Vicinity of Madagascar's Mantadia National Park. Ambio,
26, 180-184.
Simpson, R.D., Sedjo, R.A., & Reid, J.W. (1996) Valuing biodiversity for use in
pharmaceutical research. Journal of Political Economy, 104, 163-185.
Tisdell, C., Wilson, C., & Nantha, H.S. (2005) Policies for saving a rare Australian
glider: economics and ecology. Biological Conservation, 123, 237-248.
Tol, R.S.J. (2005) The marginal damage costs of carbon dioxide emissions:an
assessment of the uncertainties. Energy Policy, 33, 2064–2074.
Turpie, J.K. (2003) The existence value of biodiversity in South Africa: how interest,
experience, knowledge, income and perceived level of threat influence local
willingness to pay. Ecological Economics, 46, 199-216.
UNDP (2005) Human Development Report 2005. UNDP, New York.
van Beukering, P.J.H., Cesar, H.S.J., & Janssen, M.A. (2003) Economic valuation
of the Leuser National Park on Sumatra, Indonesia. Ecological Economics, 44, 43-
62.
van Kooten, G.C., Eagle, A.J., Manley, J., & Smolak, T. (2004) How costly are
carbon offsets? A meta-analysis of carbon forest sinks. Environmental Science &
Policy, 7, 239-251.
Verissimo, A., Barreto, P., Mattos, M., Tarifa, R., & Uhl, C. (1992) Logging Impacts
and Prospects for Sustainable Forest Management in an Old Amazonian Frontier -
the Case of Paragominas. Forest Ecology and Management, 55, 169-199.
Weitzman M (2001) Gamma Discounting, American Economic Review, 91, 260-271.
21

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Appendix 1: Full Methods
This Appendix describes in detail the methods used in the cost-benefit analysis. It
summarises the assumptions made, and the sources used in the calculation. The full,
original calculations are also available, on request, from the authors.
Currencies, Time Horizon and Net Present Values
All figures are presented in 2005 US$. Dollar values taken from other studies were
converted to 2005 dollars using a Consumer Price Index-based deflator, provided by
the Bureau of Labor Statistics (http://www.bls.gov/cpi/). Unless otherwise stated by
the authors of a study, we take all values to be in the year prior to the publication
date of that study.
Values in other currencies (FMG, Ariary or Euros) were converted to US$ using the
exchange rate at the time of the relevant study (mid-year, median, inter-bank
exchange rate as quoted by FX Currency Converter
(http://www.oanda.com/convert/classic). These dollar values are then converted to
2005 dollars as above.
We present all costs and benefits as Net Present Values (NPVs), using a discount
rate of 5% (see below) and a time horizon of 60 years. We discuss the effects of
changing the discount rate or time horizon above. The time horizon is similar to,
though slightly longer than that normally adopted for such appraisals (see for
example Dixon et al. 1986) reflecting the long term nature of the scenarios – logging
rotations are evaluated over 30-60 years for example.
Social Discount Rate
Choosing the Social Discount Rate (SDR) to use in a Social Cost-Benefit Analysis is
extremely difficult. The SDR used in a study determines the relative weight given to
communal consumption at different points in time (Kula 2004). The rate used can
have a significant effect on both the quantitative and qualitative results of a study.
Numerous authors now regard the social time preference rate, rather than the
opportunity cost of public investment, to be the principal justification for discounting in
social cost-benefit analyses (e.g. Kula 2004, see also Price 1993 for a full
discussion). This is particularly true of projects with significant non-market or non-
financial benefits and costs, and inter-generational implications. This view is not
confined to academics – since 2003, for example, the UK treasury has required all
government projects, including overseas assistance, to be discounted using an
appropriate SDR based on social time preference rate (HM Treasury 2003).
The SDR should be appropriate for the community affected by the project. In our
case, we are concerned with several communities: local people living close to the
forest corridor; the rest of Madagascar; and the rest of the world, in particular
residents of high income OECD countries. For simplicity, and because of the limited
data available, we considered two communities: Madagascar, and the high income
OECD. Unfortunately, though estimates of SDR for high income countries are
becoming more common, very few exist for developing countries. To our knowledge,
the estimates of India’s SDR as 5.2% by Kula (2004) and 2% by Sharma et al. (1991)
remain the only published estimates.
The Social Discount Rate is composed of two components. First, a rate of pure time
preference (p) which reflects the tendency for people to prefer consumption now
rather than later, regardless of their expectation of changes in per capita
I

Appendix 1: Full Methods
consumption. The second component reflects the diminishing marginal utility of
income, as income rises. It is composed of the per capita growth in consumption (g),
and the elasticity of the marginal utility of income, e.
These are combined according to the following formula:
SDR = (e x g) + p
(c.f. HM Treasury 2003, Kula 2004)
Pure Time Preference Rate p
HM Treasury (2003) interprets this as being composed of two sub-components,
“Catastrophe risk” and “Pure time preference”. Catastrophe risk is the risk of a
catastrophic event, such as natural disaster, which eliminates all returns from the
policies under evaluation. This can also include positive developments, such as
technological advancement, which render the policy obsolete. Pure time preference
represents “individuals’ preference for consumption now, rather than later, with an
unchanging level of consumption per capita over time” HM Treasury (2003). Kula
(2004) on the other hand, interprets this as being the mortality rate, m, with an
individual’s expected mortality providing the justification for pure time preference.
For all practical purposes, the distinction between these two interpretations is not
important, since estimates of the two factors for developed countries do not differ
widely, and no estimate exists for the former for developing countries.
HM Treasury (2003) reports estimates of p of 1.5, 1.3 and between 1.0 and 1.6 (Scott
[1977] Scott [1989] OXERA [2002] respectively, quoted in HM Treasury [2003]). Kula
(2004) taking p to be the mortality rate, uses 1.3 for the period 1965-1995 for India.
The current mortality rate13 for Madagascar is around 1.3 (WHO 200614). The various
figures are summarised in Table A1 below.
Table A1: Estimates of the pure rate of time preference.
Madagascar High Income OECD
Estimate Type Source Estimate Type Source
1.21 Mortality rate, reflecting
expected increase in life
expectancy
1 Mortality rate Evans (2005) for
High Income
OECD
1.3 Mortality Rate WHO
2006
1.3 Pure Time
Preference, for
UK/USA
Scott (1989)
1.32
(1.0-1.6)
Pure Time
Preference, for
UK/USA
OXERA (2002)
for UK
1.53 Rate of time
preference.
HM
Treasury
(2003)
1.5 Pure Time
Preference, for
UK.
Scott (1977).
Adopted by HM
Treasury (2003)
Notes on Table: 1Mortality rates would be expected to fall in Madagascar in the future, as they have in
the past, though this may depend on the rate of economic growth. 2Midpoint value, range reported by
the authors is given in brackets. 3This value could be justified by arguing that a developing country
would not be expected to have a pure rate of time preference less than that considered suitable for the
UK. This of course presupposes that the UK rate is suitable.
13 WHO publish age-standardized death rates, to allow comparisons between countries with different
age structures. However, the economically relevant death rate is the uncorrected one. There is only a
small difference between uncorrected rates for developing and developed countries, since the former
tend to have relatively more young people, reducing their death rates despite poorer public health.
14
http://www.who.int/ncd_surveillance/infobase/web/InfoBasePolicyMaker/CountryProfiles/QuickCompare.
aspx?DM=10&Countries=450&Year=2002&sf1=mo.cg.990&Sex=all
II

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Growth of Per Capita Consumption g
The growth rate of per capita consumption in real terms (g) is highly problematic, for
two reasons. The first problem is one of statistical availability: no reliable statistics
are available on consumer spending for Madagascar (c.f. Kula [2004], Evans [2005]).
Therefore we need to use some proxy for these, such as gross domestic product
(GDP) or Gross National Income (GNI). Even more difficult, we need to predict the
growth rate over the next 40-60 years. Even for a country which has shown relatively
stable economic growth over the last 40 years, such as the UK, this would be a
formidable challenge. In Madagascar, on the other hand, growth has been patchy
and largely negative: annual GDP per capita growth between 1975 and 2003 was
-1.6%, and that between 1990 and 2003 was -0.9% (UNDP 2005). Extrapolating
these rates into the future would lead to a negative discount rate! However, for the
last five years or so growth rates (however measured) have been better, but still
highly unstable: ranging from -15.1% in 2002, to 6.8% in 2003. Ignoring these
years15, the average has been around 2.5%.
Faced with such uncertainty, it is probably sensible to take a broader view, and
consider the average performance, past and future, of developing countries. Average
annual growth rate in GDP per capita, between 1975 and 2003 has been 2.3%
across all developing countries (UNDP 2005). Sub-saharan Africa has averaged -
0.7%, while the rate for East Asia and the Pacific, is much higher, at 6%. The World
Bank (200616) forecasts sub-Saharan Africa to grow (real GDP per capita) with an
average annual rate of 3.2% per year between 2004 and 2008.
For Madagascar, therefore, it seems reasonable to adopt an upper figure of 3.2%
and a lower figure of 2.3%, while bearing in mind the possibility that it could be
considerably lower. The High-Income OECD grew at an average of 2.2% p.a. from
1975-2003. World Bank predictions are for continued growth at an average 2.5% per
year between 2004 and 2008 (World Bank 2006).
Table A2: Estimates of per capita growth rates.
Madagascar High Income OECD
Estimate Region and
Period
Source Estimate Type Source
-1.6% Madagascar,
1975-2003
UNDP
2005
2.2% High
Income
OECD
1975-2003
UNDP 2005
2.3% Developing
countries,
1975-2003
UNDP
2005
3.2 Sub-Saharan
Africa
2004-2008
World
Bank
2006
2.5% 2004-2008 World Bank 2006
15 The very low figure for 2002 is due to the political and social crisis which gripped the country during
this year, while the very high figure for 2003 obviously reflects a partial bounce back by the economy
from an artificially low level. Over the five year period the economy still shrank by 2.5%. It is tempting to
regard such events as exceptional, but a glance at Madagascar’s modern history makes this seem
unwise. A five year period is a suitable period over which to evaluate the effects of such political crises,
tied as they are to presidential elections, which occur every 4-5 years.
16
http://web.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTDECPROSPECTS/EXTGBLPROSPECTS
APRIL/0,,menuPK:659178~pagePK:64218926~piPK:64218953~theSitePK:659149,00.html
III

Appendix 1: Full Methods
Elasticity of Marginal Utility of Consumption (e)
This is perhaps the most difficult parameter to estimate, and again, while estimates
for developing countries are relatively numerous, those for developing countries are
very rare.
It is often assumed that elasticity is constant across all incomes, and therefore may
be the same across all countries, but evidence for this is hardly overpowering, with
little empirical support for constant elasticity (though see Blue & Tweeten 1997).
Kula (2004) and Sharma (1991) provide the only estimates of e for developing
countries, while HM Treasury (2003) recommends using unity. Evans (2005)
suggests the appropriate value for developed countries may be close to 1.4.
Table A3: Estimates of e the elasticity of marginal utility of consumption
Madagascar High Income OECD
Estimate Region and
Period
Source Estimate Source
1 UK HM
Treasury
(2003)
0.95
(0.8-1.1)
UK OXERA (2002)
1.4 India Sharma
et al
(1991)
1.1
(0.7-1.5)
UK Pearce & Ulph
(1995)17
1.64 India Kula
(2004)
1.4 Hi
Income
OECD
Evans (2005)
Note: Where authors give a range, the midpoint is quoted, with the range in brackets.
Estimates of the Social Discount Rate
Using the various estimates given in Tables A1-A3 above, we calculate upper, lower
and medium values for the SDR (Table A4, below).
Table A4: Estimates of social discount rates (SDR)
Community High Incom OECD Madagascar
Value Low Med High Low Med High
p 1.00 1.30 1.50 1.20 1.30 1.50
g 2.20 2.35 2.50 -1.60 2.30 3.20
e 0.95 1.10 1.40 1.64* 1.40 1.64
SDR 3.09 3.89 5.00 -1.42 4.52 6.75
*Because of the negative value for g, the highest value for e gives the lowest value for the SDR.
In principle, the rate for each community should be applied separately to their
respective costs and benefits: the Cost-Benefit Analysis would therefore have more
than one discount rate.
In practice, given the high degree of uncertainty associated with the estimates, and
the fact the OECD estimate is entirely contained within the Madagascar estimate, we
have used a single discount rate (5%) throughout the appraisal, but have carried out
sensitivity analyses using rates from 3 to 10%.
17 Cited in HM Treasury (2003)
IV

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Why is the rate adopted so low?
Many people used to conventional economic appraisal, particularly from the last
century, may be surprised at the choice of such a low principal discount rate (5%) in
this study, even though it is towards the upper range of reasonable estimates18. It is
true that in the past organisations like the World Bank, as well as finance ministries,
used much higher discount rates, particularly in the developing world. The choice of a
lower rate here reflects both a change in the basis for the discount rate (no longer
based on opportunity cost of capital) but also more realistic estimates of growth rates
over sustained periods.
In addition, there has perhaps been a persistent misunderstanding of the relationship
between poverty and discount rates. It is often contended, usually without evidence,
that the poor must have higher discount rates than the rich. Yet proper reflection on
the reasons for discounting shows this to be untrue. Although poor people have
higher mortality rates than the rich, the difference is small (of the order of 0.2-0.5%).
The principal basis for discounting, whether one considers social discount rates or
financial discount rates, is some expectation that “life will get better”, or that
investments will bear fruit. In SDR, this is represented by the term “eg”. Where the
rate of improvement in life (g) is multiplied by the effect of this improvement on the
value of income (e). If one is chronically poor, like most rural Malagasy, one can only
expect life to get better gradually, if at all, and there is no reason why we would
expect a chronically poor person to have a high discount rate. Indeed, in the
“extreme” case of zero or negative growth, (which has been the reality for many
developing countries over the last 30 years) one can expect very low or even
negative discount rates. This is in sharp contrast to someone who is acutely poor.
For example, an otherwise comfortable person stricken by a famine or other disaster
can reasonably expect things to get much better, very quickly, if they can only survive
the present. Such a person may be expected to have a relatively high discount rate,
just like any other person who expects their income to increase dramatically: the
discount rate is independent of the starting point. The truth is that for most of the
world’s poor, life is less dramatic: they have a relatively low chance of dying in any
given year, and can expect things to be pretty much the same in the future, as they
have been in the past. The absence of any state funded welfare safety net in most
developing countries may also lower people’s discount rates. Moseley (2001)
provides empirical evidence for low discount rates among poor people in Africa, and
counsels against presuming high discount rates when looking for explanations of
environmental conversion (e.g. deforestation) by the poor.
In fact, the rate used is likely to be on the high side, for many reasons:
Firstly, it is important to note that some authors (e.g. Price 1993) have questioned
discounting on the basis of mortality or pure time preference, for social projects, and
therefore the inclusion of the mortality rate in the SDR. Excluding it would lower the
rate by just over 1%.
Secondly, the rate chosen is almost certainly too high for some sectors of the
population, including High Income OECD beneficiaries, and, perhaps, those poorest
members of society whose income may be expected to grow at a lower rate than
better placed members of society.
Finally, HM Treasury (2003, Annex 6) state that when projects are evaluated over
periods longer than 30 years, the discount rate used should fall over time:
18 The use of much higher rates (10%) in the sensitivity analysis, is designed, in part, to address the
concerns of this group.
V

Appendix 1: Full Methods
“The main rationale for declining long-term discount rates results from uncertainty
about the future. This uncertainty can be shown to cause declining discount rates
over time” Weitzman (2001).
Population and Household Size
We use commune population figures from the ILO census (Minten et al. 2003).
Unfortunately, population figures are not easily available below the commune level.
The proportion of the corridor communes’ populations living in fokontany bordering
the forest was estimated from those communes for which we were able to collect
fokontany population figures at first hand. We then calculated a weighted average of
these proportions and applied it to the total corridor population to give a corridor wide
estimate (35%, based on figures from five communes).
An estimate of mean household size is extremely important since the analysis often
required the multiplication of figures estimated “per household” by overall population
estimates. Unfortunately, the number of people in a household depends on the
definition of a household, and sources do not always specify their mean household
size. We have two estimates of mean household size in the corridor. Ferraro (1994)
calculated a household size of six persons, while information received from several
communes indicates an average of 8.5 persons. For calculations involving only
Ferraro’s (1994) estimates, we use Ferraro’s estimate of household size. For all other
calculations we used the average (weighted by sample size) of all estimates (8.3).
Extent of “The Corridor” and Scope of the Analysis
In this analysis the corridor has been taken to run from Ranomafana National Park in
the north, to Andringitra National Park and Pic d’Ivohibe Special Reserve in the
south. However, we structured the analysis to allow it to be easily extended further
north or south as necessary, although this would require collecting additional local
data. In fact, most of the analysis could easily be applied to other protected areas in
Madagascar, using locally relevant parameters.
We have not included in the analysis the low altitude forest fragments to the east of
the corridor, identified by the biological priority-setting workshop (Alliance
Ecorégionale [2005]). It is unlikely that the inclusion of these would greatly affect any
of the results. Nor have we extended the analysis fully to cover the proposed
southern extension of the corridor, to Vondrozo. However we do consider the
increase in global biodiversity benefits from conserving this area.
Forest Cover and Deforestation
Deforestation Rates
Deforestation between 1990 and 2000 was approximately 1.12% per year, across all
altitude classes (MIARO 2005). Under scenario A, the null scenario, we have
assumed that deforestation will continue at this rate, for the rest of the period
considered by this study (60 years).
This assumption may be unsafe, for two reasons. Firstly, Green and Sussman (1990)
found that in high population density areas (>5 people per km2) of the eastern
rainforests; deforestation rates fell from 2.5% per year between 1950-1973 to 0.79%
per year during 1973-1985, because suitable land for clearance gradually became
exhausted. Most corridor areas are high density according to Green and Sussman’s
study and it is possible that deforestation rates would fall as suitable land becomes
limiting.
VI

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Secondly, it must be noted that the deforestation estimate we use was calculated
over a period when considerable foreign funded activity was taking place in the
region with the aim of reducing deforestation. Hawkins and Horning (2001) suggest
this activity may have succeeded in achieving this aim. In that case, the “real”
deforestation rate, in the absence of conservation interventions, may be much higher.
These two potential biases would tend to act in opposite directions and it is beyond
the scope of this study to inquire in greater depth into the driving forces of
deforestation rates and to predict a true, real deforestation rate for the next 60 years.
We therefore use the figure of 1.12%.
Forest Cover
In order to derive an estimate of forest cover in 2005, we use the MIARO (2005)
estimate of forest cover in 2000 (384,104 ha), and the deforestation rate (see above),
and extrapolate using this deforestation rate to 2005, to give an estimated cover for
the end of 2005 of 359,107 ha.
In order to estimate the area of forest that would be included in a strictly protected
core zone, we have used a surface of natural forest cover created by PACT in 2000,
and created buffers of 1,2,3,4 and 5km from the forest edge using ArcView 3.3 (ESRI
2002). We then calculated the proportion of forest in each of these bands, and
applied these proportions to the estimated forest cover for 2005, to give an up to date
estimate of the area of forest likely to be found in each band. These areas are used
when calculating the opportunity costs, management costs or benefits arising from
the forest under strict protection (see below).
Non-Timber Forest Product Collection
The value of forest product collection to households
No estimate exists for the value of non-timber forest product19 (NTFP) collection
along the full length of the corridor. The most suitable estimate comes from Ferraro
(1994, 2002), who estimated the opportunity cost of lost forest product collection
rights20 to local communities surrounding the then newly established Ranomafana
National Park. He found these to be approximately US$ 26 per household per year
(Ferraro 1994, p80). We have used this as an estimate of the total net benefits of
forest product collection from primary forest within the Conservation Site, and applied
it to households in Forest Frontier Fokontany along the whole of the corridor.
At the April workshop, we were able to collect a great deal of qualitative data from a
number of communities, which we used to determine whether the types of NTFPs
collected and the systems of collection, use and commercialisation seemed similar to
that described by Ferraro. We concluded that the corridor as a whole appeared to be
similar to Ranomafana and that it would be reasonable to use Ferraro’s estimates as
an approximation for the whole corridor. This may nevertheless be an underestimate
of the value of NTFP collection from the Conservation Site since in the case of
Ranomafana, not all primary forest was included in the park, whereas in the case of
the corridor, we assume that all primary forest is included within the Conservation
Site.
19 We have included in this category the harvesting of wood for subsistence use, but not the extraction
of timber for commercial reasons, which is dealt with below.
20 Including traditional “rights” which may not be recognised in law, but were nonetheless believed by
local people.
VII

Appendix 1: Full Methods
Ferraro makes no estimate of the effect of losing forest to agriculture, nor does he
take account of substitutions which residents may be able to make away from forest
products (Ferraro 1994, pp 39-40). Because these are crucial to determining the
effect of the conservation site, we attempt to take account of these effects (see
below). Ferraro’s estimates are, however, net benefits (subtracting labour costs) so
we make no further adjustment for this.
In extrapolating from Ferraro (1994), we have deliberately used estimates of NTFP
value per household, rather than per hectare. We believe this to be much more
accurate (see for example Chomitz & Kumari 1998 p28) because, very often, much
of the forest is under-utilised for NTFPs, and it is the number of people living around
the forest and available markets rather than the quantity of forest, which determines
the quantity and value of NTFPs extracted.
The opportunity costs of the Conservation Site
In the case of Ranomafana, all NTFP collection was prohibited within the park, and
therefore residents were expected to lose all access to this forest. In the case of a
Conservation Site in the corridor, we have assumed that only a part of the forest
would be designated as a strictly protected core zone in which NTFP collection would
be prohibited21.
Both scenario A and B allow some NTFP harvesting therefore. The difference is that
under scenario A, forest at the edge is gradually lost to tavy, which may reduce the
amount of NTFPs that can be harvested. In Scenario B, however, no forest is lost to
tavy (tending to increase the value of NTFP harvesting relative to A) yet some forest
in the centre of the corridor is lost to the strictly protected core zone (having the
opposite effect). The net benefits of B over A depend therefore on the relative areas
lost, and the way in which the loss affects households’ collection of NTFPs. The
opportunity costs in each case will differ, per hectare, as we discuss below, but
information about the spatial extent of villagers’ forest use is essential for determining
them in both cases – we need to understand where in the forest households harvest
their NTFPs.
Spatial extent of NTFP harvesting
Unfortunately, little good information is available on the spatial extent of villagers’
forest use. Unpublished data (Vokatry ny Ala) for 2 villages suggests that most forest
use takes place within 4-5km of the village, with 3rd quartile distance of activities for
one village equal to 2.2 km (Figure A1 next page).
21 Except for the existing protected areas, which are treated as being completely strictly protected.
Households living in the periphery of these areas are treated as having lost all access to primary forest
for NTFP collection under scenario B.
VIII

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Figure A1. Spatial extent of NTFP collection
Cumulative Proportion of Forest Use with increasing distance from the Village
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Distance from Village (m)
Cumulative Proportion of Forest Use
Source: Vokatry ny Ala (unpublished data)
Opportunity cost of losing forest to the Strictly Protected Core Zone
The opportunity cost of losing forest areas to the strictly protected core is valued at
0.5 times the value of the products that would have come from that area, as
estimated using spatial information of NTFP collection (Figure A1, above [Vokatry ny
Ala, unpublished data]). This assumes that forest products, and the revenues derived
from them are at least partially replaceable with products and activities outside the
forest, and that this rate of replacement is approximately 50%. In fact, we lack good
empirical information on this rate of substitution. We estimate these opportunity costs
for five sizes of Core Zone, to allow us to assess the effect on household opportunity
costs of increasing the size of the Core Zone.
Opportunity Costs of losing forest to Deforestation
Just as we have no quantitative information on the rate of substitution for products
and revenues lost to the strictly protected core zone, we cannot quantify the rate of
substitution for products lost to deforestation. However, it seems reasonable to
assume that the opportunity cost of losing a hectare of forest to deforestation, is less
than losing a hectare of forest to protection, and may be very low indeed. This is true
for two reasons.
Firstly, when farmers clear a hectare of forest it is replaced; at various times during
the cultivation / fallow cycle, with vegetation which supplies at least some forest
products, or close substitutes. For example, roofing can be made from bamboo
collected in the forest, but it can also be made from bamboo growing outside the
forest or from thatch (tegnina or bozaka) collected from degraded lands outside the
forest.
The second reason is that along most of the corridor, the forest is not fully utilised for
forest products – most products are collected from near the edge of the forest. As the
edge moves back, forest is lost, but people move with the edge. Even though some
people may lose access to the forest, the number of people living next to the forest
IX

Appendix 1: Full Methods
can be expected to remain roughly constant. In much of the corridor, forest is being
cleared by the offspring of villagers living next to the forest, not by complete
outsiders. In this case the total benefit derived from the forest, in terms of forest
products, by those villagers (including their offspring), is likely to remain relatively
constant.
Using our estimates of spatial patterns of forest use, we see that beyond 5km from
the forest edge, forest is essentially unused. There are 6300 ha of forest 5km from
the forest edge. If we assume that forest use moves with the frontier, we find that
each hectare lost from the outer band, can be replaced by a hectare from the next
band, and so on. The first 6300 ha of deforestation are essentially “free”, with no cost
in terms of lost opportunities for harvesting NTFPs born by local people. The next
12,228 ha are priced at the value per hectare of forest products collected within the
band 4-5km from the forest edge, multiplied by 0.3. This factor of 0.3 represents the
assumption that most forest products have substitutes outside the forest, which are
at least 70% as good. Note this rate of substitution (70%) is higher than that for the
Strictly Protected Core Zone,
Affected Population
Ferraro’s estimates apply to villages in the periphery of Ranomafana National Park,
i.e. within 3km of the park boundaries. We have assumed these criteria to be roughly
equivalent to our category of Forest Frontier Fokontany, and have therefore applied
these estimates to our estimated population for the Forest Frontier Fokontany. Note
that Ferraro’s estimates do not take account of population growth. This seems
reasonable given our assumption that the frontier would move with population
growth, essentially maintaining the number of people living on the frontier
approximately constant.
Tavy
Upper Estimate: Ferraro (1994)
Only one estimate exists of the opportunity costs of stopping tavy in the eastern
rainforests of Madagascar: that of Ferraro (1994, 2002). He made estimates for the
villages surrounding Ranomafana National Park, on a per-household basis. We have
used these estimates, and multiplied them by the number of households in Forest
Frontier Fokontany (using Ferraro’s average household size). This gives the upper
estimate of the opportunity costs of stopping tavy. In the same way as for NTFP
costs, we have assumed that our category of Forest Frontier Fokontany corresponds
to Ferraro’s peripheral villages.
In fact, the per capita deforestation figures reported by Ferraro (1994) were lower
than those calculated by MIARO (2005) for the corridor, by a factor of 3-10, though
the Miaro estimate was still similar to that reported by Green & Sussman (1990) as
shown in Table A5 below. This indicates that the upper limit of the opportunity costs
of stopping tavy could be higher, if the amount of tavy which would be stopped in the
corridor, per capita, was higher than in the case studies by Ferraro. In addition,
Ferraro’s deforestation rates actually refer to forest that would have been cleared by
each household within the periphery of the park: therefore, one would expect that his
estimates would be underestimates of the total amount of forest cleared by
households. However, it would be inappropriate simply to multiply opportunity costs
calculated per-household by a correction factor to account for the possibly higher
deforestation rates in the corridor - it is possible that Ferraro’s deforestation rates,
based as they are on local knowledge, rather than remote sensing, may be more
accurate (Sader 1995).
X

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Table A5: A comparison of per capita deforestation rates
Miaro (2005) Ferraro (1994)
Using
Forest
Fokontany
Population
Using
Forest
Commune
Population
Upper Lower
Green &
Sussman
(1990)
Per Capita
Deforestation
Rate
0.033 0.012 0.006 0.001 0.024
Sources: Green & Sussman (1990), Ferraro (1994), Miaro (2005).
It should also be noted that Ferraro’s calculations of opportunity cost include
population growth rate and a model of declining fertility, and that his NPVs are for
each household and its descendents. Thus, they cannot easily be annualised.
However, the NPV per household that we give in most of our results is the same as
that given by Ferraro – i.e. using a 60 year time horizon and 5% discount rate. The
annualised figures are only used to allow us to change the time horizon and discount
rate to look at sensitivity overall.
Lower Estimate: Kremen et al. (2000)
Kremen et al. (2000), published estimates of the opportunity costs of tavy rice
agriculture, per hectare deforested, for the Masoala Peninsula. We have used these
estimates, and applied them to corridor specific figures for deforestation rates, to give
corridor specific estimates. This provides the lower estimates of the opportunity
costs.
ANGAP Management costs
We assume that the strictly protected core of the site de conservation will be
managed by an organisation similar to ANGAP22 (Madagascar’s national parks
authority) and that this organisation would have similar costs per hectare. Carret and
Loyer (2003) report ANGAP management costs to be approximately $3 per hectare,
so total management costs are taken to be equal to this figure (in 2005 dollars)
multiplied by the area of the three existing protected areas, plus the area of the
strictly protected core of the site de conservation. We assume that the buffer zone
would be managed by the communities surrounding the corridor, though we are
unable to make any estimate of the management costs of community forest
management associations (CoBas)
Since ANGAP is currently almost entirely funded by foreign aid, these management
costs represent a cost at the international level, and a benefit at the national level (a
transfer overall). In the future, donors expect ANGAP to be self-financing, and these
costs may be transferred to the national level. This will only be possible if benefits at
the national level increase, for example through capturing the value of carbon
sequestration or through increased ecotourism revenues.
22 ANGAP = Association National pour la Gestion des Aires Protégées
XI

Appendix 1: Full Methods
Ecotourism
There is currently no large-scale ecotourism in the main body of the corridor i.e.
outside of the national parks. One small community-run reserve (Anja) and a private
reserve (Ialatsara) also get significant visitors but these are not contiguous with the
corridor. We have information from the National Parks and Anja reserve on the direct
benefits, such as entrance fees.
We have based our estimates of the direct benefits of ecotourism on these figures,
using a correction factor of 20% to account for the fact that we have incomplete
coverage of the corridor. We have used estimates of the indirect benefits of
ecotourism, quoted in Carret and Loyer (2003), and estimates for the distribution of
those benefits between local, regional and national levels, given in ANGAP (No
Date). Carret & Loyer (2003) give 55$ per person indirect benefits from ecotourism
and this is partitioned as per fig 2 in ANGAP (no date). We assume that there is
some overlap between Ranomafana, Anja and Andringitra and use this to decrease
the national indirect benefits accordingly.
“Local” revenues from ANGAP (no date) were divided into 1/5 for Forest Frontier
Fokontany and 4/5 for Corridor Communes following Peters (1998) who estimated
that only 1/5 of local tourism revenue went to people resident within the park’s
periphery prior to 1989 – broadly equivalent to our category of Forest Frontier
Fokontany.
We assume that ecotourism in the corridor as a whole will grow at between 8 and
10% per year, following the World Tourism Organisation projections for the Indian
Ocean (http://www.world-tourism.org/), for the next 25 years, and then at 4-5%. We
also assume that the composition of that tourism, in terms of the distribution of
benefits remains approximately the same. The overall benefits are likely to be
broadly accurate, though the distribution of benefits will depend on the way tourism
develops.
We lack estimates of the management costs of the Anja reserve. Some of these
management costs will be in the form of cash spent in the community, and will thus
probably not decrease net benefits. Some will be disbursed farther afield. The
inclusion of management costs would therefore likely reduce net benefits slightly, and
push benefits out from the fokontany to commune, regional and national levels.
We assume that any corridor ecotourism has no benefit at the international level,
since it plays a negligibly small part in world tourism meaning that it has near-perfect
substitutes, i.e. the presence of the corridor as an ecotourism venue has no net
benefit to international ecotourists. The international value of the corridor is however,
taken account of through its value for carbon sequestration, bio-prospecting and non-
use values (see below).
We have used 2004 figures for ANGAP ecotourism (ANGAP unpublished data).
These may change in 2005 because entrance fees have been substantially
increased, but without knowing the price elasticity of demand it is impossible to
predict the effect of these changes.
This still assumes that there is no substitutability between the corridor and other sites
in Madagascar, in terms of ecotourism venues – i.e. that if the corridor was lost, no
tourists would switch to other Madagascar destinations, they would simply be “lost” to
Madagascar.
We have presumed ecotourism to be completely incompatible with Scenario A.
Overall, these assumptions are likely to be generous to the case for conservation,
and perhaps to overestimate its local benefits.
XII

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Bio-prospecting
It is notoriously difficult to determine the value of genetic diversity and many early
estimates appear to have been wildly optimistic (Simpson et al. 1996). However, the
eastern rainforests have an extremely high degree of endemism, and therefore may
be relatively valuable per hectare. Having reviewed the literature on bio-prospecting
values, van Beukering et al. (2003) use $1 per hectare for a tropical rainforest park in
Indonesia, and we followed them in using this value.
For illustrative purposes we have taken the value of forest outside of the strictly
protected core to have 75% the value of forests inside. For scenario A, we have used
the expected area of forest at the end of the 40 years, for scenario B; we have used
the estimated area of forest in 2005.
At present, we have assumed that pharmaceutical values, like those of carbon
sequestration and the non-use values of biodiversity accrue only at the international
level, because currently no mechanisms are in place to transfer them to local levels.
Note that the local value of medicinal plants is captured in estimates of NTFP use by
villagers (Ferraro 1994, 2002).
Hydrological Benefits of Forest Protection
Over the last two decades, our understanding of the effect of deforestation on
hydrological function has advanced considerably. Two major reviews by Bruijnzeel
(1990, 2004) have collected and synthesised a large number of empirical and
modelling studies on the subject, while Chomitz & Kumari (1996, 1998) have
reviewed the economic implications of deforestation, reforestation and forest
protection. Meanwhile, Brand et al. (2002) have reviewed much of this from a
Malagasy perspective. Appendix 3 summarises these findings.
The conclusions of these reviews are that many of the previously claimed benefits of
tropical forests for hydrology have either been overstated, or are too uncertain to
include in valuations. In particular, Bruijnzeel (1990, 2004) stresses that
deforestation, per se, is not the cause of many hydrological problems, but rather that
they are the result of poor management of land after it has been cleared. Many of the
ways in which deforestation can cause hydrological problems are associated with
heavy logging operations, which may not be relevant in the corridor, and can in any
case be greatly alleviated by using appropriate techniques.
Nevertheless, it remains the case that the protection of forests will likely have a
benefit for those living and farming downstream, particularly in terms of avoided
erosion. We have used the study by Brand et al. (2002), of the willingness to pay
(WTP) of downstream rice farmers to stop deforestation up stream, as an estimate of
this. This study found that 50% of farmers were willing to pay to prevent deforestation
up stream, and that for these farmers, average WTP was US$2 per household, at the
time of the study. We took this value and applied it to 50% of all agricultural
households in the Forest Frontier Fokontany and Communes. The fact that only 50%
of farmers were willing to pay perhaps reflects the fact that some farmers perceived
sedimentation to be a benefit rather than a cost, a view which the authors appear to
endorse. It is possible therefore that these farmers would have been WTP to ensure
upstream deforestation, and that in applying this figure we overestimate the benefits
of forest protection.
Obviously, there are great uncertainties in transferring such an estimate to the
corridor – in particular, Brand et al. (2002) do not provide much information on their
study area. At the very least, we would have liked to use both their WTP per
XIII

Appendix 1: Full Methods
household, and per hectare of irrigated rice paddy, but we lack data on the number of
hectares of irrigated rice paddy surrounding the corridor.
Non-use Values of Biodiversity to non-Malagasy
Citizens of rich countries have repeatedly been found to be willing to fund nature
conservation, in tropical countries, for reasons unrelated to tangible benefits they
hope to gain from those resources. Numerous Contingent Valuation (CV) studies
have been conducted on residents of rich countries to determine their willingness to
pay (WTP) to preserve this or that habitat. However, none has yet been conducted
concerning Madagascar, except on ecotourists already in Madagascar. In the
absence of a corridor specific study, one possible approach would be to use a
“basket” of valuation studies, each of which has valued a nature conservation site
that was “similar” to the corridor. Some average of their valuations could then be
used to estimate the value that rich world citizens might place on the continued
existence of the corridor, and be willing to pay for its preservation. This approach has
been termed “benefits transfer” (Bateman et al. 2002). Clearly, this would be
imperfect, and heavily dependent on what was classed as “similar” to the corridor.
In addition, valuations of single sites have been criticised on the grounds that
respondents in CV studies attribute all of their willingness to pay for conservation in
general, to whichever specific site or species they are asked about in the study. Thus
the CV reveals not the WTP for the site in question, but for all wildlife sites (see for
example Blamey 1996, Kahneman & Knetsch 1992). Price (No Date, p16) provides a
good summary of this, and other such problems.
For this reason, it is probably more realistic to use estimates of the WTP for
conservation in general, and estimate the share of that WTP which is attributable to
the corridor. Unfortunately, such aggregate studies are rare. We found just two
suitable published studies. Kramer & Mercer (1997) investigated the WTP of US
citizens for rainforest conservation, whereas Menzel (2005) investigated German
citizens WTP for biodiversity conservation.
Kramer & Mercer (1997): WTP for tropical rainforest conservation
Kramer & Mercer (1997) used CV to discover the one-time willingness to pay of US
citizens to conserve an extra 5% of tropical rainforests throughout the world (over
and above the 5% already in protected areas). The corridor would appear to fit this
category, since it represents an extension to Madagascar’s protected area network,
and is one of the highest priority conservation areas in the world (Mittermeier et al.
2004).
We have assumed that rest of rich world (European Economic Area, Canada, Japan,
Australia, New Zealand) would exhibit similar WTP to the US. We calculated average
per capita WTP from Kramer and Mercer’s per household estimate ($11.9), then
multiplied this by rich world population of 868m in 2004 (OECD 2004). This total WTP
for extra tropical rainforest conservation was then assumed to be shared out equally
amongst 5% of all the world’s tropical forests, on a per hectare basis. The corridor
accounts for 0.73% of this additional 5% of the world’s tropical rainforests. The
global, one-time willingness to pay for conservation of the corridor, based on the
Kramer and Mercer study is just over $78m.
This estimate is likely to be conservative, for two reasons:
1) Awareness of rainforests, and biodiversity has increased substantially since
the study was conducted in 1992.
XIV

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
2) We have only included the richest OECD countries in the study: namely the
EU15, Norway, Iceland, Switzerland, US, Canada, Japan, Australia and New
Zealand. This ignores any value placed on the Malagasy forests by residents
of other countries, in particular the substantial number of rich citizens of
Eastern Europe, South East Asia and South and Central America.
The study is likely to represent only non-use values. Although the study may capture
some use-values, the authors believed this to be low: the study was carried out
before climate change became a major issue, and before international ecotourism
was common place making it unlikely that respondents willingness to pay includes a
great deal of use value.
Menzel (2005): WTP for biodiversity conservation
Menzel (2005) used CV to estimate the monthly contribution which German residents
over the age of 18 were prepared to make, in the form of a “biodiversity tax” for the
“protection of half of the endangered species expected to become extinct in the next
10 years” (Menzel 2005, p33). The study asked respondents to reply for themselves,
and not for their family or household, yielding a per capita value. The mean response
was approximately €9 per month. This yields an annual figure, for the whole of
Germany of just over €7bn per annum. However, the author states that this could be
an overestimate, because people may, despite instructions to the contrary, have
replied on behalf of their household. If we assume that the WTP is therefore a per
household WTP, the result becomes €3.8bn ($4.5bn) per annum.
Using the per household estimate as a conservative estimate, we multiply by the
population ratio of the rich world to Germany (868m:82.5m) to give an annual
extrapolated total WTP for the rich world of $47.4bn.
To interpret this result in the context of the corridor, we have taken this WTP for the
protection of half of the species facing extinction as a conservative estimate of the
WTP to conserve all of these species. To calculate the corridor’s contribution to such
an aim, we have used the percentage of the critically endangered bird and mammal
species worldwide which are endemic to the corridor. There are 362 critically
endangered birds and mammals worldwide (IUCN, quoted in Mittermeier et al. 2004)
and three are endemic to the corridor (Randrianasolo pers. com.). Thus, by this
reasoning, conservation of the corridor would achieve 0.8% of the worldwide task of
preserving all critically endangered species. This is likely to be an underestimate of
the corridor’s contribution, since it ignores those critically endangered species which
are found elsewhere in Madagascar as well as in the corridor - the corridor would still
play an important part in their conservation. However, using this assumption, we
arrive at a global, annual WTP for the conservation of the corridor of $393m. This
estimate may also underestimate WTP for conservation because it may not include
other aspects of conservation which people value, including landscape and
wilderness values, as well as ignoring the values of people outside of the rich world.
Unlike Kramer & Mercer’s study, Menzel aimed to elicit an annual, ongoing WTP. If
we assume that this WTP would be repeated for each of the 40 year time horizon of
the study, this would give a NPV of $6.7 trillion. However, for the purposes of this
study, we have taken the conservative assumption that the value elicited by Menzel
was a one off WTP, and therefore used $393m.
It should be noted that Menzel did not adjust for the age bias in her figures, which
would lead to a small upward bias in her estimates, however, this should be more
than compensated-for by taking her annual figure to be a one-time figure.
XV

Appendix 1: Full Methods
Vondrozo Extension
The international non-use value of the Vondrozo extension was calculated in the
same way as above. The value based on Kramer & Mercer (1997) was calculated
using the estimated additional hectares of the extension (73,808 as estimated by the
authors). The value based on Menzel (2005) was calculated using the fact that the
Vondrozo extension adds at least one extra endemic mammal or bird species:
Eulemur albocollaris, which is found only in a small portion of the corridor south of
Andringitra, except for a small, very isolated population in Manombo Special
Reserve, near Farafangana (Irwin et al. 2005).
Timber
Information on timber production was provided by a specialist group at the workshop
consisting of Eaux et Forêts personnel, commercial loggers, and regional
development workers. This information was cross-checked with other specialists, and
a consensus found.
Types of logging operation
Two approaches were identified. The first involved extracting only category 2 and 3
timber species that have reached a given diameter. In principle, this would allow the
forest to be logged, on a rotation basis, returning every 10-15 years (Rasamisandy
pers com.) This rotation length seems quite short, compared to practices elsewhere
(e.g. 30-40 years [Verissimo et al. 1992] 60 years [Healey et al. 2000], but without
more information on the productivity of corridor forests, and the diameter criteria
used, it is difficult to confirm this. Nevertheless we have used a longer rotation length
of 30 years, meaning that two rotations would take place within the time period
considered by the study. In this approach, Category 4 timbers, would not be
extracted at all, and only half of Category 3 timbers, making this approach relatively
conservative on timber volumes.
In the second approach, the forest would be logged for all useable timber, regardless
of size and species (though not including prohibited Category 1 species). Participants
agreed that this production was feasible and economic, but would mean that the
forest could not be logged again for a long time – it is essentially a one-time
extraction, and would be represent a greater initial impact on the forest.
XVI

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Total value of production
For each approach, experts estimated the total production per hectare, and its value
on the open market (table A6 below).
Table A6. Timber production per hectare for two different methods.
Scenario Cat
m3
harvested
/ ha
Product Pieces
/ m3
Pieces
per
hectare
price per
piece on
market
Total Value
of
Production
(Ar) per ha
2 3 1 56 56 6,000 336,000
3 3 1 56 56 4,000 224,000
4 0 0 0 0 0 0
Multiple
rotation
Total (Ar) / ha 560,000
2 7.2 2.4 56 134 7,143 960,000
3 9.69 3.23 56 181 3,118 564,000
4 9.69 3.23 56 181 3,118 564,000
One-time
Total (Ar) / ha 2,088,000
Source: Eaux et Forets, Workshop Participants
In order to calculate the net present value of each approach, we first used the
extrapolation of deforestation estimates (see above) to determine the amount of
forest available in each year. All forest in the corridor was estimated to be exploitable
(workshop participants, Rasamisandy, pers. coms.).
For the first, “multiple rotation23” approach, we assumed that in any given year,
logging takes place in a portion of the forest remaining in that year equal to 1/r
(where r is the rotation length). The gross value of production was given by the
number of hectares multiplied by the per hectare value given in table A6 above.
For the second, “one-time” approach, we assumed that the whole of the corridor is
logged within the first 25 years, and is not returned to within the time horizon of this
study (60 years). In each of the first 25 years, 1/25th of the forest is logged.
These annual, gross, market values of production are then discounted to give a
Gross Present Value of production.
Distribution of benefits
The gross value of production is divided into benefits for the state (through taxes)
labour (through wages), costs (principally transport) leaving a remainder for the
owner of the logging operation. The type of logging practiced in the corridor is not
capital intensive, and we assumed that all except transport costs represent a net
benefit to Madagascar, and we justify this below.
Taxes
In order to calculate the distribution of this product among different stakeholder
groups, we first deducted the various taxes from the total gross value of production.
These taxes represent a benefit to the state at various levels, as shown in table A7
below.
23 We have deliberately avoided the term “sustainable” here.
XVII

Appendix 1: Full Methods
Table A7: Taxes on logging operations
Percentage Beneficiary Stakeholder Level
12% Eaux et
Forets
Nation
8% Fiscal Nation
3% Ristournes Fokontany, Commune, Region (see below)
Ristournes assumed to be divided in the following way:
10% (0.3% of total) Fokontany Forest Frontier Fokontany
30% (0.9% of total) Commune Forest Frontier Commune
30% (0.9%) District Corridor Regions
30% (0.9%) Region Corridor Regions
Implicitly we have assumed that logging operations pay all of their taxes, on the full
value of their production. If this assumption was violated, it would of course reduce
the benefit to state, but increase that to private individuals. Since most logging
operations are owned by Malagasy nationals, this would not have a large effect on
the distribution of benefits among the stakeholder groups considered in this analysis.
Labour
Next we used estimates of the labour costs required to cut the wood and carry it to
the road (Rasamisandy pers. com.) to determine the benefit to labour. These are
shown in table A8 below.
Table A8 Labour costs in logging operations
Approach Cat
Saw
Wages
per
piece
Porter
Wages
per
Piece
Piece per
hectare
Total Saw
wage per ha
(Ar)
Total Porter
wage per ha (Ar)
2 1800 1,150 56 100,800 64,400
3 1800 900 56 100,800 50,400
Multiple
rotation Total 201,600 114,800
2 1800 1,150 134 241,920 154,560
3 1800 900 180.88 325,584 162,792
4 100 900 180.88 18,088 162,792
One-time
Total 585,592 480,144
Source: (Rasamisandy pers. com.)
100% of the portering wages are assumed to accrue to residents of forest fokontany,
while only 25% of the saw team wages are; the remainder are assumed to accrue
residents of the communes, outside of the fokontany (see Ferraro 1994 pD-32).
These gross benefits are assumed to be approximately equal to net benefits,
because we assume that the opportunity cost of labour is low in these areas, with few
alternative sources of employment outside of the agricultural season.
Transport
Average transport costs were estimated at 300 Ar per plank, for delivery from the
road head to Fianarantsoa. These come to 33,600 Ar per hectare for the multiple
XVIII

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
rotation scenario, and 148,848 Ar for the one-time scenario. Transport costs are
subtracted from the total production, and are not considered to be benefits.
Logging Concessionaire
The remainder, after labour, transport and taxes are taken into account, is the net
benefit for the concessionaire. Smaller concessionaires may be residents of corridor
communes, while larger businesses may be owned by those from farther a field. We
have assumed that concessionaires’ benefits are divided 20:50:30 among the
communes, regions and nation. The breakdown of benefits, per hectare, is given in
Table A9 below.
Table A9: Breakdown of benefits from logging operations
"Multiple Rotation" "One-time"
Share Value (Ar) per Ha Share Value (Ar) per Ha
Saw Workers 36% 201,600 28% 585,592
Porters 21% 114,800 23% 480,144
E&F 12% 67,200 12% 250,560
Ristournes 3% 16,800 3% 62,640
Fisc 8% 44,800 8% 167,040
Exploitant 15% 81,200 19% 393,176
Transport 6% 33,600 7% 148,848
Logging within the Conservation Site
We evaluated the potential economic benefits, in terms of timber exploitation, of
allowing communities to log, sustainably, a portion of the corridor outside of the
strictly protected central zone. Using estimates of the area outside the central zone
(see above), we calculate the gross present value of logging using the same total
production per hectare as the multiple rotation approach above, but assuming a
rotation length of 40 years instead of 30 for increased sustainability and lower impact.
We have assumed that benefits will be distributed as for Scenario A, in other words
we do not presume that the villages retain any greater percentage of the value of
production. If villages did manage to do so, e.g. by developing skills in saw work, or
even by fulfilling the role of concessionaires, this would increase the local benefits of
this scenario.
It must be noted, that we have not attempted to evaluate any of the costs of this
scenario, compared to scenario B. This is because most of these, such as the impact
of logging on biodiversity or ecotourism are impossible to judge (although the effects
on hydrological function are likely to be negligible (Bruijnzeel 1990, 2004, Serpantié
pers com). It should be possible to determine the effect on carbon sequestration (see
for example Healey et al. 2000) but this would require a more precise understanding
of the exact methods to be used in logging.
Carbon
The sequestration of carbon dioxide, the main greenhouse gas, is one of the most
important functions fulfilled by tropical forests and avoiding deforestation avoids the
emission of CO2. Using estimates of forest cover, and deforestation rate published by
MIARO (2005), we estimate the number of hectares of forest which would be lost
under scenario A, and which are therefore saved by the Conservation Site. van
Kooten et al. (2004) review studies of carbon offsets, and we use their mean value
across all studies to give an upper estimate of the CO2 released by a conversion of a
hectare of tropical forest (220 tonnes per ha). The lower estimate comes from
XIX

Appendix 1: Full Methods
Razafindralambo’s estimate for the Zahamena-Mantadia corridor (148 t/ha [R.
Razafindralambo, unpublished data]).
There are several ways to value the benefits of avoiding the release of CO2. The first
is to use the price of the CO2 on the world market. This represents the value which
might be captured directly by Madagascar, through the Kyoto protocol. The second is
to use estimates of the real marginal cost of CO2 emissions. Although the
uncertainties in this estimate are obviously much greater, this represents the “true”
value of avoided CO2 emissions. We use the latter (estimated by Tol 2005) to
calculate the benefit to the global community of preventing deforestation in the
corridor. We use his median estimate of $14.42 as our lower estimate, and his mean
estimate of $44.29 as our upper estimate. These values are then multiplied by the
quantity of CO2 release avoided each year, and the NPV calculated.
Cultural or non-use values of the corridor to Malagasy
As well as citizens of rich countries Malagasy residents themselves may hold non-
use or cultural values for the Malagasy forests. Unfortunately, we know of no attempt
to determine the benefit which Malagasy residents gain simply from the continued
existence of forests. Indeed, such assessments, while common in the developed
world, are comparatively rare in developing countries. Those which have been
carried out, have often demonstrated a significant willingness to pay for conservation
among urban citizens (Hadker et al. [1997], Turpie [2003]).
One response to the absence of a specific study which identified Malagasy citizens’
WTP for corridor conservation would be to take a benefits transfer approach,
whereby the results of other studies were applied to the context of the corridor, and
Madagascar. However given the sparse literature on the subject, and the absence of
any studies for Madagascar, we do not feel this to be an option at this stage.
In absolute terms, we might expect this WTP to be modest: incomes in Madagascar
are low compared with, say South Africa, where Turpie (2003) found substantial WTP
which increased with income. The lowest annual household income group in her
study was <$1200, which is quite high in the Malagasy context. In addition, there
appear to be low rates of forest-based recreation by Malagasy citizens, in contrast to
the South Africa case. However, where this does occur (e.g. Anja reserve) it would
be interesting to conduct a revealed preference study.
Nevertheless, if one considers the cultural and traditional importance of forests to the
Malagasy, particularly those residing close to the forest, these may be significant. In
an attempt to address this issue, in qualitative terms, and to evaluate what it may
mean for the Conservation Site, a session of the workshop was devoted to
discussing cultural and traditional values which residents of the corridor hold. This
session uncovered a wealth of information about sites of cultural importance,
traditional mechanisms for the management of forests, and the attitudes of the focus
group towards outside interventions. In general, the group noted that any planning
process for the Conservation Site should take account of the traditional and cultural
importance of forest sites. Access to areas of special cultural importance, for
example those containing tombs or which provide products such as medicinal plants
should be guaranteed. In addition, activities, such as ecotourism or timber
exploitation which might conflict with some of these sites, should be controlled
carefully. The group exhibited a distrust of government or outside management of
forests, while finding no conflict per se between conservation of the forest and
traditional values. If implemented sensitively, the Conservation Site should only
provide benefits to local populations in terms of non-use values.
XX

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
It is difficult to generalise across the whole region, and it must be noted that only a
small and not necessarily representative group was present. However, the main
lesson is that the Conservation Site need not conflict with local non-use values, but
that this is contingent on it being implemented sensitively at the local level, with
adequate participation in the planning process by local people. In terms of the
analysis, it is likely that the inclusion of local non-use values of the forest in the
economic evaluation would decrease the net costs of the Conservation Site for local
people. However, we do not believe that these values would be sufficient to reverse
the sign of the net benefits at the local level. If they were large enough to do this, it is
likely that local conservation measures would have rendered government protection
of the forest unnecessary. It seems more likely that local measures alone would
result in the conservation of patches of forest of particular significance, but not the
whole corridor. This retention of “sacred forests” is certainly seen in other areas of
Madagascar, which are otherwise denuded of forest.
XXI

Appendix 2: Full results table
Table A10: Full breakdown of results, using 60 year time horizon and 5% discount rate.
For uncalculated benefits, we have given our opinion of the likely qualitative effect that including the benefit would have on the value of the Conservation Site
(+: positive; -: negative; =: no significant effect.) In the final column, we have estimated the direct financial benefits of allowing low impact logging in 1/3 of the
conservation site. We have noted the likely effect of this on other values with the same symbols.
Scenario A - Null Scenario
Scenario B - Site de conservation with
no logging and Strictly Protected Core
at 3km
Net benefit of B over A
Benefits of low impact
logging in 1/3 of buffer
zone of Scenario B
Beneficiary group Benefit Lower Upper Lower Upper Lower Mid Range Upper
Forest Frontier Fokontany Non-Timber Forest products 14,670,491 14,670,491 12,382,055 12,382,055 -2,288,436 -2,288,436 -2,288,436 =
(FFF) Tavy 5,928,452 15,908,600 0 0 -15,908,600 -10,918,526 -5,928,452 =
Ecotourism 0 0 10,003,046 16,293,514 10,003,046 13,148,280 16,293,514 -
Irrigation 0 0 860,320 860,320 860,320 860,320 860,320 =
Timber 13,062,372 47,474,802 0 0 -47,474,802 -30,268,587 -13,062,372 2,784,423
FFF Net Benefit -54,808,473 -29,466,950 -4,125,426
Uncalculated benefits Through Travel -/= =
Mining 0 0 - - - =
Cultural Values / Non-use
values + -
Drinking Water =/+ -
Forest Frontier Communes Ecotourism 0 0 6,762,553 11,015,221 6,762,553 8,888,887 11,015,221 -
(FFC) Irrigation 0 0 911,409 911,409 911,409 911,409 911,409 =
Timber 13,500,706 40,258,760 0 0 -40,258,760 -26,879,733 -13,500,706 2,877,860
FFC Net Benefit (excl FFF) -32,584,798 -17,079,437 -1,574,076
FFC Net Benefit (Inc FFF i.e. cumulative) -87,393,271 -46,546,387 -5,699,502
Uncalculated benefits Through Travel -/= =
Mining --- =
Cultural Values + -
Drinking Water =/+ -

Region Ecotourism 0 0 12,711,196 20,704,699 12,711,196 16,707,947 20,704,699 -
Timber 3,966,928 17,568,211 0 0 -17,568,211 -10,767,570 -3,966,928 845,605
Region Net Benefit (excl FFC & FFF) -4,857,015 5,940,378 16,737,770
Region Net Benefit (Inc FFC & FFF i.e. cumulative) -92,250,286 -40,606,009 11,038,268
Uncalculated benefits Mining --- =
Irrigation =/+ =
Non-use values + -
Drinking Water =/+ =
National Ecotourism 0 0 54,612,090 88,955,193 54,612,090 71,783,641 88,955,193 -
Timber 10,673,448 40,178,606 0 0 -40,178,606 -25,426,027 -10,673,448 2,275,191
Management costs 0 0 7,031,328 7,031,328 7,031,328 7,031,328 7,031,328 =
National Net Benefit (excl Corridor Regions) 21,464,811 53,388,942 85,313,073
Total National Net Benefit (i.e. cumulative) -70,785,475 12,782,933 96,351,341
Uncalculated benefits Mining --- =
Non-use values + -
External Carbon Sequestration 0 0 29,654,109 135,595,037 29,654,109 82,624,573 135,595,037 =/-
Bioprospecting 0 129,594 109,049 109,049 -20,545 44,252 109,049 -
Non-use values 0 0 63,230,420 410,069,946 63,230,420 236,650,183 410,069,946 -
Management Costs 0 0 -7,031,328 -7,031,328 -7,031,328 -7,031,328 -7,031,328 =
External Net Benefit 92,863,985 319,319,009 545,774,032
Total Net Benefit
Madagascar -70,785,475 12,782,933 96,351,341
Total Net Benefit World 22,078,510 332,101,942 642,125,374

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Appendix 3: Notes on Hydrology and Tropical
Deforestation
Compiled by Jonathan E Annis1 and Neal J. Hockley2
1 Department of Civil and Environmental Engineering, Michigan Technological
University, Houghton, MI USA. jonathanannis@mac.com.
2 School of the Environment & Natural Resources, University of Wales, Bangor.
Deiniol Road, Bangor. LL57 2UW. United Kingdom. neal.hockley@univ.bangor.ac.uk.
Summary
In this Appendix, we summarise the state of knowledge about the hydrological
functions of tropical forests, the effects of deforestation and the economic
significance of these effects.
While tropical forests have some important hydrological functions that may be
adversely affected by deforestation, the hydrological importance of deforestation
has often been overestimated, and consequently the economic value of forests
that is due to their hydrological function has sometimes been overstated. In many
cases where the received wisdom asserted that forests were essential for
maintaining hydrological function, empirical and modelling studies have
suggested that the effects were smaller than previously assumed. In some cases,
even the direction of the effect was shown to be inconsistent, or the reverse of
that previously assumed. Many hydrological functions performed by forests have
been shown to be compatible with other land uses.
The evidence from hydrological science therefore points firstly to a reduction in
the probable size of the hydrological effects of deforestation, and secondly to an
increase in the uncertainty over its magnitude, and direction. When translated into
economic terms, this means that the value of the hydrological functions of tropical
forests is likely to be much lower than previously assumed.
Hydrological Implications of Tropical Deforestation
The following section aims to summarise two major reviews of the hydrological
implications of deforestation in the tropics written by L. A Bruijnzeel (199024, and
2004)25. To help the reader verify what we have written, statements are followed by
the page number of the original document, for example: “(1990:188)” means
Bruijnzeel (1990, page 188). Statements in quotation marks have been copied
verbatim from the text (though we have removed the citations). For the full references
please see the original documents. Suggestions, interpretations or comments that we
have made relative to the specific situation of the corridor or Madagascar are marked
with “g”.
We have attempted to present a balanced review of the most important points. If our
review appears to be sceptical about the hydrological importance of tropical forests,
this reflects the thrust of the original reviews and the fact that the received wisdom
has up until now been overly optimistic about it. It is worth quoting from the foreword
24 Bruijnzeel (1990). Hydrology of Moist Tropical Forests and Effects of Conversion: A State of
Knowledge Review published by the UNESCO International Hydrological Programme.
25 Bruijnzeel (2004). Hydrological functions of tropical forests: not seeing the soil for the trees?
Agriculture, Ecosystems and Environment 104 p185–228.
XXIV

Appendix 3: Notes on Hydrology and Deforestation. J. E. Annis & N. J. Hockley.
to Bruijnzeel’s 1990 review, written by Lawrence S. Hamilton, of the Environment and
Policy Institute, East-West Center in Hawaii:
“In spite of [attempts] to dispel the misinformation,
misinterpretation, misunderstanding and myth about the
role of forests with regard to hydrology and erosion, and
what happens when the forest is altered or removed, many
of these “four Ms” still continue to dominate popular and
political thinking. This applies especially to tropical
humid forests, which seem to automatically put emotion
into command over reason”
Effect of Forest Clearance on Rainfall/Climactic Conditions
 “Apart from forests at specific locations, such as coastal fog belts or cloud belts in
mountainous areas, or forests of very large areal extent (e.g. the Amazon basin),
tropical forests most probably do not influence local amounts of rainfall
significantly” (1990:179).
 “Given the right conditions, increases in total precipitation through cloud stripping
can be sizeable, especially so for isolated single trees or rows of trees. The effect is
somewhat less in the case of closed forest due to the mutual sheltering of trees but
hundreds of mm yr-1 may be contributed by occult precipitation in forests subjected
to persistent wind-driven fog and/or clouds. Typical values for the wet tropics range
between 4 and 18 per cent of ordinary rainfall to over 100 per cent under more
seasonal conditions”. (1990:15)
 The few empirical studies which have been conducted show mixed results, with
some even demonstrating an increase in rainfall after deforestation. Major
modelling studies of deforestation in the Amazon show that even deforestation of
the entire area might result in only an 18% reduction in rainfall26. In the case of
isolated forest corridors the effect will likely be insignificant.
 “The impact of land cover on the precipitation signal is expected to be muted in
regions with a large oceanic contribution, such as southeast Asia and the Pacific,
West Africa, the Caribbean side of Central America and north-western South
America” (2004:188).
Summary
gThe Ranomafana – Andringitra corridor area appears to exhibit several
characteristics (oceanic climate; thin, isolated corridor, relatively high rainfall), that
would tend to suggest that forest cover is not the principal driver of rainfall patterns,
though in some of the higher parts of the corridor, cloud forests may be important in
increasing local rainfall. However, the finding that cloud stripping is particularly
pronounced when trees occur in isolation, or in thin bands, rather than as closed
forest, may mean that maintaining the large areas of forest demanded by
conservation is not necessary to maintain precipitation levels.
26 Lean, J. and P.R. Rowntree (1993). A GCM simulation of the impact of Amazonian deforestation on
climate using an improved canopy representation. Quarterly Journal of the Royal Meteorological
Society, 119, 509-530.
XXV

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Effect of forests on water yield.
Effect of Forest Clearance on Total Water Yield
 Forests do not act as “sponges” soaking up water during the rainy season and
releasing it during the dry season. Rather, they act more like pumps that continually
uptake water from the substrate (2004:187, 194-195).
 “Although forest soils generally have higher infiltration and storage capacities than
soils with less organic matter often much of this water is consumed again by the
forest rather than used to sustain streamflow.” (1990:79)
 Removal of forest cover leads to higher stream flow totals and reforestation of open
lands generally leads to a decline in overall stream flow. This is due to the
decreased evaporative loss from tall vegetation during rainy spells (interception)
and by the lack of transpiration from trees during dry periods (2004:201-209).
 “The bulk of the increase in flow upon clearing is normally observed in the form of
baseflow, as long as the intake capacity of the surface soil is not impaired too
much” (2004:195).
 “With respect to the influence of forests on water yield (total streamflow) it is
beyond doubt that both natural and (mature) manmade forests use more water
than most agricultural crops or grass land”. (1990:179).
 “Generally, the initial increases in total water yield following forest clearing exhibit a
more or less irregular decline to pre-clearing levels with time, reflecting the
development of the regenerating or newly planted vegetation and year-to-year
variability in rainfall” (2004:195).
 “The results presented thus far all pertain to relatively small catchment areas
[usually <1 km] involving a unidirectional change in cover. Although these provide a
clear and consistent picture of increased water yield following a replacement of tall
vegetation by a shorter one and vice versa, effects of conversion may be more
difficult to discern in the case of larger basins having a variety of land use types
and vegetation in various stages of regeneration.” (1990:96-97)
Effect of Forest Clearance on Seasonal Distribution of Flows
 “In many parts of the tropics, especially in areas experiencing a dry season, the
seasonal distribution of streamflow assumes greater importance than total annual
water yield.” (1990:98)
 “Reports of greatly diminished dry season flow abound in the literature and are
usually ascribed to “deforestation”.... At first sight this would seem to contradict the
evidence presented [above] in respect to increases in total water yield following
removal of tall vegetation… However, the conflict can be resolved when taking into
account the net effects of changes in ET and infiltration opportunities associated
with the respective land use types… Summarising, if infiltration opportunities after
forest removal have decreased to the extent that the increase in amounts of water
leaving the area as stormflow exceeds the gain in baseflow associated with
decreased ET, then diminished dry season flow is the result.” (1990:111).
 “the continued exposure of bare soil after forest clearance to intense rainfall, the
compaction of topsoil by machinery or overgrazing the gradual disappearance of
soil faunal activity and the increases in area occupied by impervious surfaces such
as roads and settlements all contribute to gradually reduced rainfall infiltration
opportunities in cleared areas. As a result, catchment response to rainfall becomes
more pronounced and the increases in storm runoff during the rainy season may
become so large as to seriously impair the recharging of the soil and groundwater
XXVI

Appendix 3: Notes on Hydrology and Deforestation. J. E. Annis & N. J. Hockley.
reserves feedings springs and maintaining baseflow… When this critical stage is
reached, diminished dry season (or ‘minimum’) flows inevitably follow”. (2004:201)
 “This situation, of course, is widespread in the tropics and can generally be held
responsible for the deterioration of streamflow regimes so commonly
observed”.(1990:111)
 “the commonly observed deterioration in river regimes following tropical-forest
removal is not so much the result of the clearing itself but rather reflects a lack of
good land husbandry during and after the operation… this is precisely where our
hope for the future lies”. (1990:113)
Reforestation and water yields
 “Although reforestation and soil conservation measures can reduce enhanced peak
flows and stormflows associated with soil degradation, there is no well-documented
case of a corresponding increase in low flows. While this may reflect higher water
use of newly planted trees, cumulative soil erosion during the post-clearing phase
may have reduced soil water storage opportunities too much for remediation to
have a net positive effect in particularly bad cases.” (2004:217)
 “water yields have been reported to return to original levels within 8 years where
pine plantations replaced natural forest, such as in upland Kenya”. (2004:198)
 The water use of eucalypts “on soils of intermediate depth (ca. 3m) was not
significantly different from that of indigenous dry deciduous forest However, on
much deeper soils (>8 m) the annual water use of the plantations exceeded annual
rainfall considerably, suggesting ‘mining’ of soil water reserves that had
accumulated previously in deeper layers during years of above-average rainfall.
(2004:199)
 Therefore: “planting of eucalypts, particularly in sub-humid climates, should
therefore be based on judicious planning, i.e. away from water courses and
depressions or wherever the roots would have rapid access to groundwater
reserves”. (2004:199)
g Summary: Deforestation may have little effect on total stream flow, but may
decrease dry season flow if the land husbandry which follows deforestation is poor.
In the corridor, rainfall is seasonal, and dry season flows are important for irrigated
rice farming. The replacement of forest by overgrazed or regularly burnt pasture,
such as is seen frequently on the western side, may well lead to a reduction in
minimum flows. Replacing forest with perennial crops, or fallow cycle agriculture, on
the eastern side, may not have much effect. Reforestation, may be able to restore
minimum water yields, but care should be taken over the choice of species and site.
Effect of Forest Clearance on Flooding
 “The hydrological response of small catchment areas to rainfall (stormflow
production) depends on the interplay between climatic, geological and land use
variables. Key parameters in this respect include the hydraulic conductivity of the
soil at different depths, rainfall intensity and duration, and slope
morphology.” (2004:203)
 “…a certain increase in stormflow volume and peakflow magnitude cannot be
avoided after forest conversion, even with minimal surface disturbance”. (1990:103)
 The type of clearing is important factor in determining storm flow response of a
disturbed forest watershed. (2004:203-204)
XXVII

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
 “[Studies] indicate the strong (local) increases in stormflow volume and peakflow
that may be brought about by such adverse practices as regular burning or
overgrazing of grassland or forest undergrowth”. (1990:103)
 “Normally, peaks (and to a lesser extent stormflow volumes) produced by some
form of overland flow are more pronounced than those generated by subsurface
types of flow. Therefore, the dramatic increases in peakflows/stormflows that are
often reported after logging or land clearing operations using heavy machinery
primarily reflect a shift from subsurface flow to overland flow dominated stormflow
patterns as a result of increased soil compaction.” (2004:204)
 “in catchments where overland flow (usually of the ‘saturation’ type) is already
rampant under undisturbed conditions (for instance due to the presence of an
impeding layer at shallow depth; the response to rainfall after forest removal hardly
increases any further. An example of soils with such poor hydraulic conductivity,
and thus high surface runoff even under forested conditions in southeast Asia are
the shallow heavy clay soils developed from marls in Java”. (2004:204)
g It would be interested to know how the heavy, clay soils of the naturally forested
corridor respond, especially given that the bedrock is only just below the surface in
many places, and that they therefore would appear to meet these criteria?
 “it is important to not immediately attribute short-term trends in the frequency of
occurrence of peak discharges or floods on large river systems to upstream
changes in land use.” (2004:205)
 “Truly widespread flooding is usually the result of an equally large field of extreme
rain, occurring at a time when soils have become wetted up by previous rains. In
such cases, the process of runoff generation is governed by soil water storage
capacity rather than topsoil infiltration opportunities. The presence or absence of a
well-developed vegetation cover has become of minor importance by then.” (1990:180)
 “the measure often used for policy making is the economic loss associated with a
particular flood… equating economic losses with severity of flooding may introduce
serious bias in that it gives the impression that flooding has become more frequent
and damaging in recent years, whilst in reality the increased economic losses
mainly reflect economic growth and increased floodplain occupancy.” (1990:108)
 “Whilst it is beyond doubt that adverse land use practices after forest clearance
cause serious increases in stormflow volumes and peakflows, one has to be careful
to extrapolate such local effects to larger areas. High stormflows generated by
heavy rain on a misused part of a river basin may be ‘diluted’ by more modest flows
from other parts receiving less or no rainfall at the time, or having regenerating
vegetation”. (2004:205)
g Summary: In general then, poor land husbandry in the corridor could cause
increased flooding at the scale of the fokontany or commune, but will likely have little
effect at the regional level.
Effects of Forest Clearance on Soil Erosion
 “in steep terrain, it is rather difficult to evaluate the influence of various
disturbances on land-slide frequency and magnitude with any degree of certainty.
Also, remaining areas of forest in the tropics are often found on slopes too steep for
terraced cultivation. This immediately introduces the methodological difficulty of
finding comparable control sites (the forested slopes being steeper and therefore
more susceptible to gravity). In addition, it is not uncommon that such breaks in
slope reflect a change in lithology as well”. (1990-122)
XXVIII

Appendix 3: Notes on Hydrology and Deforestation. J. E. Annis & N. J. Hockley.
“it is helpful to distinguish between surface erosion, gully erosion, and mass
movements, because the ability of a vegetation cover to control these various forms
of erosion is rather different.” (2004:209)
Surface Erosion
 “This form of erosion is rarely significant in areas where the soil surface is protected
against the direct impact of the rain, be it through a litter layer maintained by some
sort of vegetation or through the application of a mulching layer in an agricultural
context.” (2004:211)
 “Erosion … becomes considerable upon repeated disturbance of the soil by
burning, frequent weeding or overgrazing, which all tend to make the soil
compacted or crusted, with impaired infiltration and accelerated erosion as a
result”. (2004::211)
Gully erosion
 “Gully erosion is a relatively rare phenomenon in most rain forests but may be
triggered during extreme rainfall when the soil becomes exposed through treefall or
landslips”. (2004:212)
 “…active gullying in formerly forested areas is often related to compaction of the
soil by overgrazing or the improper discharging of runoff from roads, trails and
settlements.” (2004:212)
 “[Some studies have] stressed the importance of gullies to catchment sediment
yield in view of the increased ‘connectivity’ afforded by gullies between hillslope
fields and streams... The moderating effect of vegetation on actively eroding gullies
is limited and additional mechanical measures such as check dams, retaining walls
and diversion ditches will be needed.” (2004:212)
g There appears to be a significant difference between the eastern and western sides
of the corridor: while gullies can be observed in many areas on the western side,
they appear to be rare on the east (G. Serpantié, pers com27). This may be due to
differing soil types, lower levels of grazing pressure or the presence of better
ground cover on the eastern side.
Mass Wasting
 “[Although] the presence of a forest cover is generally considered important in the
prevention of shallow (<1m) slides, the chief factor being mechanical reinforcement
of the soil by the tree root network… Mass wasting in the form of deep-seated (>3
m) landslides is not influenced appreciably by the presence or absence of a well-
developed forest cover. Geological (degree of fracturing, seismicity), topographical
(slope steepness and shape) and climatic factors (notably rainfall) are the dominant
controls.” (2004:212)
g Summary: Once again, therefore, it is the particular land husbandry practices
which replace the forest which are important in determining the effect of deforestation
on erosion. Surface erosion and gullying will be promoted by regular burning and
overgrazing, and the eastern side of the corridor may be more resilient to on or other
of these effects.
27 Serpantié, G: IRD, Antananarivo, Madagascar. April 2005
XXIX

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
Effect of Forest Clearance on Sediment Transport
 “…any benefits of increased dry season water yield following clearing are often
more than offset by increased stream sedimentation rates”. (1990:181)
 “…applying soil conservation measures like contour cropping, bunding, grass
strips, terracing, mulching, etc. or tree planting may well reduce amounts of
sediment generated by surface erosion and shallow land slips entering the
drainage system. However, due to storage effects, it may take several decades for
basins larger than several hundred km2 before stream sediment loads in the
downstream parts will become noticeably smaller.” (1990:182)
 “As such, the frequently voiced claim that upland rehabilitation will solve most
downstream problems, does require some specification of the spatial and temporal
scales involved.” (1990:182)
g As noted by Brand et al. (2002)28, some farmers perceive a benefit from erosion
caused by deforestation, which transports soil and nutrients from upland areas to
lowland fields. Kull (2004)29 also noted this perception among farmers with respect
to the effects of burning pasture land above paddy fields.
Other notes
 When evaluating the impacts of deforestation, its extent (i.e. manual vs
mechanized clearing, conversion for agriculture/pasture lands, slash and burn
agriculture with defined fallow periods) must be well defined. (1990:68-69)
 The spatial and temporal variation in tropical rain intensities can often “drown out”
the land use effects in the analysis of large watersheds. (1990:97)
 Caution must be applied when simply comparing results for different catchments
with contrasting land uses since geological (soil characteristics) and topographical
(e.g. basin shape) factors may override the vegetation effect. (2004:195)
 “…important hydrological benefits of irrigated rice cultivation include delaying the
arrival of surface runoff peaks and trapping sediment eroded further
upslope”. (2004:192)
 “Soil moisture status is an important determinant in basin hydrological response,
with wetter conditions corresponding with a more vigorous response and vice
versa”. (1990:103)
 “The information presented … leads to the observation that the adverse
environmental conditions so often observed following “deforestation” in the humid
tropics are not so much the result of “deforestation” per se but rather of poor land
use practices after clearing of the forest. This is precisely where our hope for the
future lies.” (1990:184, original emphasis)
28 Brand, J., Healy, T., Keck, A., Minten, B., & Randrianarisoa, C. (2002). Réalités et mythes sur
l'aménagement des bassins versants : l'effet de la déforestation des pentes sur la productivité de riz
dans les plaines, Policy Brief. FOFIFA Programme ILO. August 2002. pp 5. Also published in English
as: Brand, J., Healy, T., Keck, A., Minten, B., & Randrianarisoa, C. (2002). Truths and myths in
watershed management: the effect of upland deforestation on rice productivity in the lowlands, Policy
Brief. FOFIFA Programme ILO. August 2002. pp 5.
29 Kull, C.A. (2004) Isle of fire. The political ecology of landscape burning in Madagascar University of
Chicago Press, Chicago & London.
XXX

Appendix 3: Notes on Hydrology and Deforestation. J. E. Annis & N. J. Hockley.
Economic Significance of Hydrological Functions
Two major reviews of the economic implications of the hydrological functions of
tropical forests, have been compiled by Chomitz and Kumari (199630, 199831). They
base their reviews heavily on the work of Bruijnzeel and other hydrologists (see
above). However, they have interpreted this work from an economic perspective,
looking at the domestic (within–country) hydrological benefits of preserving tropical
forests.
They stress that the hydrological effects of deforestation are far from clear, and
depend greatly on the land use that follows. Given this, if one takes a domestic point
of view, and ignores other values of the forest such as biodiversity values,
maintaining natural forest cover under strict protection will rarely be the economically
favoured way of maintaining hydrological functions of catchments. Reduced impact
logging, or the replacement of natural forest by plantations, perennial crops, or long
fallow slash and burn agriculture with small plot sizes will often be nearly as good
from a hydrological perspective as natural forest, while generating other substantial
economic benefits. In cost-benefit analyses, like that presented in the main report,
that seek to compare two options for a forest area, one based on conservation, one
on production, the hydrological benefits of maintaining some form of vegetation cover
cannot be attributed solely to the conservation option. Many of these benefits will
also be present, to some degree, in a production option that also allows greater
extractive or productive use of the land area. Evaluating the difference between two
options, in hydrological terms will be difficult, and there is no certainty that the
conservation option will significantly outperform the production option.
In addition, even if a discernible reduction in hydrological or climatic function can be
attributed to deforestation, it does not necessarily follow from this that an economic
cost will result. For example, even if deforestation could be shown to result in a
reduction in rainfall, it does not follow directly that a cost results. In areas which
typically have very high rainfall levels, such as eastern Madagascar, it is far from
clear that a reduction in rainfall would per se result in a net economic cost.
Finally one must consider the importance of the extent of any change. For while
biodiversity conservation may require large areas of undistrubed natural forest,
hydrological function may be maintained by the retention of small, possibly simplified
forests (natural or planted) in particular key areas.
Chomitz and Kumari conclude that the case for tropical forest conservation must
therefore be made substantially on other grounds, such as the value of the forest for
ecotourism, carbon sequestration, or non-use, existence-related values of its
biodiversity and landscape.
The authors state on the cover of their 1996 review that the economic benefits, in
terms of hydrological function, of tropical forests benefits are:
“likely to be highly context-specific and may often be smaller than popularly
supposed.”
30 Chomitz, K.M. & Kumari, K. (1996). The Domestic Benefits of Tropical Forests A Critical Review
Emphasizing Hydrological Functions, Policy research working paper. Rep. No. 1601. World Bank, May
1996.
31 Chomitz, K.M. & Kumari, K. (1998) The domestic benefits of tropical forests: A critical review. World
Bank Research Observer, 13, 13-35.
XXXI

Hockley & Razafindralambo (2006) A Social Cost-Benefit Analysis of Conserving the Ranomafana-Andringitra-Pic d’Ivohibe Corridor
XXXII
They go on to say that:
“This underscores the importance of grant financing to support forest
preservation that yields global or non-economic benefits.”
In other words, they recommend that we should not necessarily expect developing
countries to save tropical forests primarily because of their hydrological function, nor
should they necessarily do so. However, where tropical forests provide other
benefits, particular of a global nature, we should directly fund their conservation,
internationally, through mechanisms such as the Global Environmental Facility
(GEF)32.
Conclusion and Significance for the Corridor
Hydrologically speaking, what happens after forests are cleared depends greatly on
how the land is managed, and what vegetation replaces it. Agroforestry, plantations,
long fallow slash-and-burn or perennial crops will all perform similarly to, if less well
than, natural forest. Given the weakness of many of the linkages between
deforestation and hydrological problems, the economic implications are even less
clear, and likely to be less important than is often supposed. In the context of the
corridor, the economic implications of hydrological problems following deforestation
are far from certain, but are unlikely to be very large, and the value of un-captured
benefits in this study is unlikely to be hugely significant. It is possible, however, that
the effect may be very different on the two sides of the corridor, since land husbandry
practices and the vegetation which replaces the forest is very different. On finer
spatial scales, individual villages and communes may benefit significantly from forest
retention: to predict this would require further work. Overall then, the hydrological
benefits of the corridor should not be forgotten (nor have they been in the preceding
valuation) but neither should they be overstated or allowed to become a distraction
from the other, very significant benefits that conserving the corridor will bring to local
communities, to Madagascar, and to the world.
32 See for example Menzel, S. (2005). Financial support for biodiversity protection in developing
countries – does the CBD mechanism lead to an appropriate level of biodiversity protection? In
Valuation and Conservation of Biodiversity Interdisciplinary Perspectives on the Convention on
Biological Diversity (eds M. Markussen, R. Buse, H. Garrelts, M.A. Manez Costa, S. Menzel & R.
Marggraf), pp. xxx, 430 p. 57 illus. Springer. And other chapters in the same volume.