09/10/2012 Towards an improved methodology for the ......• Improve current (SP) valuation...
Transcript of 09/10/2012 Towards an improved methodology for the ......• Improve current (SP) valuation...
09/10/2012
Towards an improved methodology for
the valuation of ecosystem servicesThe substitution question
Jeremy De ValckPhD Student, KU LeuvenResearcher, VITO (Flemish Research Center)
09/10/2012 2© 2012, VITO NV
Content
• PhD research context
• Conceptual framework
• The substitution effect
• The case
• Further investigations
• Discussion
09/10/2012 3© 2012, VITO NV
Research context
» Drivers:
• ↗ pressure on ecosystems � ↗ demand for nature valuation from policy-
makers
• What contributes to the value of nature?
• Improve current (SP) valuation techniques: CVM, CE
» PhD research: “Valuation & mapping of ecosystem services”
• Past experience (collaboration with Roy Brouwer’s unit, IVM – VU*) showed
the need to investigate spatially-related questions: substitutes, distance-
decay, upscaling issues, etc.
• Need to account for different sources of preference heterogeneity to
improve transferability of value functions
• Final goal: Improving/generalising a valuation tool developed by VITO for
policy-makers:
http://rma.vito.be/natuurwaardeverkenner/doc/Brochure_ESD.pdf
*Cf. Liekens et al. (2013): “Developing a value function for nature development and land use policy in Flanders, Belgium”.
09/10/2012 4© 2012, VITO NV
ON-SITE
characteristics
ON-SITE
characteristics
INDIVIDUAL
characteristics
INDIVIDUAL
characteristics
OFF-SITE SPATIAL
characteristics
OFF-SITE SPATIAL
characteristicsWTPWTP
1. Recreation preferences
2. Socio-demographics
3. Others (e.g. spatial
cognitive capacities, ex
ante knowledge)
1. Ecosystem-related
2. Infrastructure-related
•Spatial context (e.g. distance to
site, number/distance to eligible
substitutes)
Conceptual framework
09/10/2012 6© 2012, VITO NV
Study area: The Drongengoed
» Context: “peri-urban” forest (456 ihb/km²)
» 860 ha one-piece nature area, made of:
• 250 ha conifers
• 310 ha broadleaves
• 25 ha heathland
• 275 ha others (pasture, arable land, peat, poplar)
» Decision on nature conversion scenarios currently at
the core of the political debate
� Traditional forestry practices Vs nature/recreation focus
09/10/2012 7© 2012, VITO NV
The survey
» 284 respondents, 252 kept for final analysis
» 11.3% “protest bidders” removed through filtering process
Variable Survey Flanders* census (Belgian Federal
Government, 2012)
Gender Male 55.8% 49.4%
Female 44.2% 50.6%
Age (years) 18 – 29 13.3% 21.5%
30 – 49 34.9% 34.3%
50+ 51.8% 44.2%
Household size (persons) 1 17.6% 12.0%
2+ 82.4% 88.0%
Level of education High school degree or lower 51.6% 72.9%
Bachelor degree or higher 48.4% 27.1%
Household net income ≤€2,500 58.9% 72.0%
>€2,500 41.1% 28.0%
Job status Employed 54.4% 66.2%
Not employed 45.6% 33.8%
Descriptive statistics of the respondents. [Source: Belgian Federal Government, 2012 (URL: http:// economie.fgov.be)]
09/10/2012 9© 2012, VITO NV
The choice experiment» Choice cards:
» 24 cards selected (out of 144) = 4*6 cards � 6 cards per respondent
» 2 alternatives + status quo
» Attributes:
1. Habitat
• Broadleaf, heathland
2. Reduction in coniferous forest
• 50ha, 100ha, 200ha
3. Biodiversity
• More common species
• More common & rare species
4. Accessibility
• Good
• Poor
5. Price (€)
• 10, 25, 50, 75, 125, 200
• Payment vehicle: annual mandatory tax exclusively used to restore/conserve the Drongengoed
09/10/2012 10© 2012, VITO NV
Results : Model I (CL) Vs Model II (MXL)
Attributes
Step 1
Model I (CL)
Step 2
Model II (MXL)Mean Std. Dev.
ASC 1.024*** 2.879***(0.167) (0.291)
Rare species 0.369*** 0.0818 3.546***(0.0950) (0.285) (0.383)
No Access -0.427*** -1.187*** 2.523***(0.0929) (0.251) (0.264)
Broadleaf -0.388*** -0.996*** 2.468***(0.143) (0.303) (0.350)
Size100 -0.663*** -0.769*** 0.551(0.107) (0.219) (0.435)
Size200 -0.391*** -1.025*** 1.951***(0.136) (0.301) (0.372)
Size100*Broadleaf 1.086*** 1.474*** 1.446***(0.204) (0.409) (0.542)
Size200*Broadleaf 0.175 0.231 1.652**(0.187) (0.468) (0.716)
Price -0.0132*** -0.0314***(0.00118) (0.00252)
Summary statisticsLog-likelihood -1464.3 -1208.3AIC (Akaike Information Criterion) 2946.6 2448.7BIC (Bayesian Information Criterion) 3004.4 2551.4Observations 4,536 4,536Sample size 252 252
Robust standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1
�Preferenceheterogeneity formost attributes
�Better fit of Model II
� Confirms the needto prefer MXL
09/10/2012 11© 2012, VITO NV
Results : Interacting attributes with additional
variablesIndividual-related variables Description
Age Respondent’s age (in years)
Gender Dummy. 1 if male
Family size Total household size in persons (adults and children)
High education Dummy. 1 if Bachelor or higher degree, 0 if High school degree or lower
Employed Dummy. 1 if currently employed
High income Dummy. 1 if income >€3,500
knowDrong Dummy. 1 if ex ante knowledge of the study site
Actual user Dummy. 1 if respondent has already visited the site
Homenat Dummy. 1 if nature proximity was crucial for choosing home location
Ecofriendly Dummy. 1 if member of an “eco-friendly” NGO (e.g. WWF)
Off-site spatial variables
Distkm Road distance (in km) between respondent’s home and the site
Natprox5km Dummy. 1 if individual feels sufficiently surrounded by nature in his 5 km vicinity
09/10/2012 12© 2012, VITO NV
Results : Ecofriendliness
COEFFICIENTS
VARIABLES Ecofriendly-interacted
ASC 0.747***
(0.185)
Rare species 0.386***
(0.106)
No Access -0.440***
(0.111)
Broadleaf -0.382**
(0.173)
Size100 -0.580***
(0.129)
Size200 -0.215
(0.163)
Size100*Broadleaf 1.042***
(0.246)
Size200*Broadleaf -0.0933
(0.222)
Price -0.0137***
(0.00123)
Ecofriendly*ASC 1.178***
(0.410)
Observations 4,500
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
�Confirms that attitude
towards nature is a crucial
parameter
�Need to better understand
people’s individual
characteristics
09/10/2012 13© 2012, VITO NV
Results : Nature proximity (5km)
VARIABLES
COEFFICIENTS
Natprox5km-
interacted
ASC 1.800***
(0.293)
Rare species 0.290*
(0.151)
No Access -0.701***
(0.160)
Broadleaf -0.186
(0.309)
Size100 -0.791***
(0.219)
Size200 -0.590***
(0.215)
Size100*Broadleaf 0.862**
(0.437)
Size200*Broadleaf 0.363
(0.340)
Price -0.0132***
(0.00120)
Natprox5km*ASC -1.046***
(0.344)
Natprox5km*No Access 0.385*
(0.197)
Observations 4,536
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
�Confirms the detrimental
effect of substitutes on
people’s WTP
�Lower WTP for nature
restoration scenarios
�Less affected by a lack of
site accesibility
09/10/2012 14© 2012, VITO NV
Results : Willingness to pay estimates
Willingness to pay estimates (€) with Krinsky-Robb 95% intervals Model I (CL) Model I – Interacted
Attributes WTP Lower limit Upper limit WTP Lower limit Upper limit
ASC 77.8 52.0 103.1 101.1 61.1 141.1
Rare species 28.0 14.9 41.6 28.9 15.1 42.6
No Access -32.4 -49.1 -18.0 -45.8 -69.9 -21.7
Broadleaf -29.4 -50.0 -8.3 -25.9 -47.0 -4.7
Size100 -50.4 -70.9 -32.8 -47.1 -65.7 -28.6
Size200 -29.7 -52.5 -9.0 -27.0 -47.7 -6.2
Size100*Broadleaf 82.5 54.0 113.9 76.3 45.0 107.7
Size200*Broadleaf 13.3 -16.0 41.5 11.0 -16.1 38.2
Additional variables interacted
Ecofriendly*ASC 91.5 46.6 136.3
Natprox5km*ASC -64.2 -106.5 -21.9
Natprox5km*No Access 23.7 -3.9 51.4
09/10/2012 15© 2012, VITO NV
Drongengoed case conclusions
» We confirm that 3 sources of preference heterogeneity impact people’s WTP for nature restoration:
1. on-site characteristics2. Individual characteristics3. off-site spatial characteristics
» People positive towards transformation of coniferous forest. (Model I: 78€/yr*HH towards heathland, 48.4€/yr*HH towards broadleaf)
» People prefer a change to more biodiversity (+28€/yr*HH), less conifers and good accessibility (-32.4€/yr*HH).
» Attitude to nature (“ecofriendliness”) is crucial. It almost doubled people’s WTP �from 101.1€ to 192.6€/yr*HH (Model II).
» Perceived substitution effect: people feeling surrounded by nature had ~60% lower WTP for nature restoration and were less affected by a lack of accessibility.����Confirms the need for further investigation of the spatial heterogeneity
question.
09/10/2012 16© 2012, VITO NV
Investigating the substitution question
» Goal
Further investigate the substitution question and create a new variable that
controls for that effect.
» Idea
Building a “Substitution effect index” (SEI) based on GIS and survey data
» Hypothesis
SEI = f(similarity, density, distance),
Where:
» similarity = capacity to supply similar level of utility to respondents
» density = abundance indicator, to be compared with the target
» distance = to account for distance-decay effect
09/10/2012 17© 2012, VITO NV
Stating the problem
» Choice decision rule based on utility maximisation: the respondent can
ascribe a level of utility to each possible destination and go for the one
maximising utility.
» Sounds like building a composite indicator by means of multi-criteria
analysis but with some tricky constraints:
» 1. Respondent-centric model � modelling preferences for one
nature site over a bunch of alternative sites (respondent-specific)
within a certain range (also respondent-specific!)
» 2. Not a pure question of recreational destinations. This is also an
amenity-related question. Both are linked.
09/10/2012 18© 2012, VITO NV
General approach
GIS-based tasks:
1. Defining the “primary site” (reference) according to a set of chosen
characteristics
2. Calculate all possible substitutes based on the same characteristics
3. Compute a “similarity index” for each substitute to indicate how
good/bad substitute it is.
Survey-based tasks:
4. Explore respondent’s answers about socio-demographic and recreational
habits.
5. Develop a weighing factor based on respondent’s preferences (e.g.
hobbies)
6. Combine the “similarity index” with the respondent-specific weighing
factor to obtain his substitutes and their related power.
09/10/2012 19© 2012, VITO NV
Primarysite
Primarysite
Individual
preferences
Individual
preferences
SusbtitutesSusbtitutes
SEI components (1)
Substitutioneffect Index
SEI Components (2)
1. Primary site (or “reference”)
Ecosystemcharacteristics
•Size
•Habitat type(s)
•Bio-physical indicators
Infrastructurecharacteristics
• Accessibility level
• Facilities
Attractiveness
• Crowdedness• NB: likely to be an individual-
dependent quadratic function (e.g. more people increase social enjoyment, but too many people decrease peacefulness)
• Points of interest
NB: Fisher et al (2009): ESS production area Vs ESS benefit area � in situ, directional or omni-directional
SEI Components (3)
2. Substitutes
Relativeproximity
• Travel cost attribute
• Ratio travel time/recreation time (individual-dependentbecause depends onrecreation preferences!)
• (Means of transport)
Direct substitutes
• Similarity level
• Number/density within a certain “range” (individual-dependent)
Othersubstitutes
• “Man-made” substitutes
• (e.g. cinema, park, theatre, amusement park, bar, museum, sport facilities)
• NB: also individual-dependent
SEI Components (4)
3. Individual-related characteristics
Recreationpreferences
• FREQUENCY
• RANGE:
• ���� Trip duration: 1h trip, day-trip, weekend?
• ���� Trip enjoyment: Some people mayenjoy the travel itself! Jones et al. (2010)
• � Depends on people’s spatialcognitive capacitites
Socio-demographics
• Source of proxies (age, gender, income, level of education, etc.)
Otheraspects
• Spatial cognitive capacities
• Ex ante knowledge
• Values
• Experience of nature
• Sentimental attachment to a type of landscape � culturalvalue
• Others
09/10/2012 24© 2012, VITO NV
Discussion (1)
Current issues:
» Hard in practice!
» GIS data availability, quality and relevance to calculate “similar”
substitutes
» Approach computationally demanding � GIS software limitations
(ArcGIS) & computing power
» Correct econometric approach? Number of observations?
Questionnaire?
09/10/2012 25© 2012, VITO NV
Discussion (2)
Current questioning:
» No distance-decay found. Conflicting with high non-use value?
» Substitutes: good approach?
» Sufficiently innovative?
» What about other spatially-
related questions (scale, δ-decay,
use><non-use)?
» Individual-related heterogeneity
(values, ex ante knowledge, beliefs,
“sense of place”, etc.)?
» Any “No way!” thus far?