M it b
Manitobans’
’D
Demand
d ffor W
Wetland
tl d
Retention and Restoration
Non--Market Valuation of Wetlands
Non
Peter Boxall
Boxall, John Pattison & Vic
Adamowicz
D
Department
t
t off Rural
R lE
Economy
University of Alberta
Wetland Valuation Challenges
• Wetlands are not homogeneous commodities
– EGS service flows are variable
– Acres versus quantities of EGSs? Are they positively
and strongly correlated?
– Human demands for services varies spatially
• What action is taken to change wetland service
flows?
– Retention? Difficult to portray – why should I pay?
– Restoration? Easier – provides increases in service
flows hence a positive environmental improvement
The Issue of Service Flow Baselines
• All valuation projects require a description of the
currentt situation
it ti
• What is this for wetlands?
–N
Need
d ffor good
d iinventory d
data
– Need to understand historical changes in wetland
quantities or changes in EGSs
• What is the future trajectory of these changes in
an absence on intervention?
– Human caused? Other factors like drought?
• Are there g
goals for interventions and how are
these determined?
Wetland Programs
• What specific actions will be taken to change
the current situation or future trajectory?
• Is there a goal? What is realistic?
• For
F this
thi study
t d action
ti iinvolves
l
th
the restoration
t ti off
wetlands
– Need a current baseline wetland inventory
– Need to understand wetland inventory in an absence
of restoration activity
– Need to understand what a budget for restoration
would look like
– These
Th
are allll significant
i ifi
t challenges
h ll
Non-Market Valuation of Wetlands
NonTh
Through
hR
Restoration
t ti iin M
Manitoba
it b
• Manitoba case study
– Issues of how much to restore, where to
restore, how much is currently there etc.
• Lots of data around, lots of controversy
– E.g.
g what is wetland loss? Presence on
landscape? Loss of function?
• What do these wetlands provide in terms
of EGS and can this be portrayed to “lay”
people?
Part of a Larger
g Research Effort
• DUC flagship watershed – Broughton Creek in SW
Manitoba – intensivelyy studied
– Prairie pothole region
– Wetlands inventory both current and historical well
studied
– Not yet extensively drained
– Dr. Wanhong Yang’s research where hydrologic model
developed to understand current EGS provision as well
as changes to EGS with restoration
• AAFC flagship watershed – South Tobacco Creek
–E
Escarpmentt area (former
(f
shore
h
off Lake
L k Agassiz)
A
i )
– More or less completely drained
– Dr. Wanhong
g Yang’s
g research where hydrologic
y
g model
developed to understand current EGS provision as well
as changes to EGS with restoration
Scope and Current Situation
• We used Broughton
g
Cr. estimates of
drainage and EGS service provision as
p
of the p
pothole region
g
in
representative
Manitoba
• Baseline chosen to be 1968 where DUC
had “good” inventory data
– 100% wetland restoration would involve
restoring to 1968 inventory levels
– Annual historic losses were about 0.57%
The Manitoba Context
B
Broughton
ht
Creek
C
k
MB Pothole Region
The Prairie Pothole Region
High level of wetland loss
•How Much?
•Since When? If When? If nothing
nothing done, what would happen?
%
DUC estimates for Broughton Cr.
• Projected loss of 7,000 acres per year (0.57% rate of loss)
Projected loss of 7 000 acres per year (0 57% rate of loss)
• Note that 1968 is the baseline year
What does decline mean?
• Pick a future year – 2020
• Assume 0.57% loss
• Translate this into loss of EGS flows for
Manitobans
• Show how this decline in EGS would change with
wetland conservation programs
– Retention of current wetland levels ((i.e. arrest the loss))
– Restore wetlands to various levels (100% restoration
would mean 1968 levels).
EGS Benefits of Wetlands
Yang et al
al. Broughton Cr
Cr.
• Water filtration:
– 0.043 tonnes/acre/yr N
– 0.009 tonnes/acre/yr
y P
•
•
•
•
Soil stabilization: 6.5 tonnes/acre/yr
Flood control: 1200 m3/acre/yr
Carbon sequestration: ~4 tonnes/acre/yr
Biodiversity: production of waterfowl (?)
How to communicate this information to a respondent?
Translated
At the current rate of loss, by 2020 wetlands will provide
the following services in the PPR of Manitoba:
• Removal of nutrients from runoff equivalent to 45 semitruck loads of fertilizer
• Provide storage for 9 million cubic meters of flood water
• Prevent loss of 50
50,000
000 tonnes of eroded soils
• Store carbon equivalent to 6,000 cars on provincial
roads
d
• Provide habitat to produce 58,000 breeding pairs of
d k
ducks
If Programs Restore XX% of
W l d that
Wetlands
h existed
i d iin 1968
1968:
by 2020 wetlands will provide the following services:
• Removal of nutrients equivalent to XX semi-truck loads
of fertilizer
• Provide storage for X million cubic meters of flood water
• Prevent loss of XX,XXX tonnes of eroded soils
• St
Store carbon
b equivalent
i l t tto X,XXX
X XXX cars on provincial
i i l
roads
• Provide habitats to produce XX,XXX breeding pairs of
ducks
But Wetlands also Impose Costs
• Opportunity Costs: Lost cultivated acreage translates
t lower
to
l
yields
i ld
• Nuisance Costs: time spent
p
driving
g around p
potholes,
increased fuel consumption, seed overlap
• Restoration Costs – unplugging drains, etc
• Crop Damage - Wildlife feeding on crops – especially
waterfowl!
• Tradeoffs – less funds available for other programs
such as health care,
care infrastructure development
development,
education, etc.
Utilized a Stated Preference
Survey
Survey
•
•
•
•
Description of baseline (trajectory)
Description of retention programs
D
Description
i ti off restoration
t ti programs
Provincial level programs – referendum
context
• “Top
p - Down” valuation approach
pp
– Not valuing the individual components of
carbon, flood control, etc., but the overall
program
Mechanisms to Address
H
Hypothetical
h i l Bi
Bias and
d other
h IIssues
• Used a cheap
p talk script
p
• Reminded respondents of budget and
substitutes
• Uncertainty scale used for the votes
– How certain are you of your vote?
– Recode
R
d U
Uncertain
t i “YES” tto “NO”
• Used debriefing questions
– why they voted for current situation; why they voted
for a proposed program
– Identification of “yea-sayers”
Pre--Briefing Script
Pre
D i IIssues: hypothetical
Design
h
th ti l bi
bias, strategic
t t i
behaviour, scope etc.
Cheap Talk Script
PLEASE NOTE: Research has shown that how people vote on a survey is often not a reliable
indication of how people would actually vote at the polls. In surveys, some people ignore the
monetary and other sacrifices they would really have to make if their vote won a majority and
became law. We call this hypothetical bias. In surveys that ask people if they would pay more for
certain services, research has found that people may say that they would pay 50% more than they
actually will in real transactions
transactions.
It is very important that you “vote” as if this were a real vote. You need to imagine that you
actually have to dig into your household budget and pay the additional costs.
Reminder of what is to follow and to treat each vote independently
•
•
•
You will now vote 5 times
Assume that the options on EACH SCREEN are the ONLY ones available
Each time, please vote independently from the other votes - do not compare options on
different screens
Valuation Scenarios
• Each respondent voted 5 times between the
current situation and a proposed restoration
program
– Referendum binary choice format
– Order of votes randomized
– Tax level in each vote randomized
• Attributes and levels describing the proposed
programs
– Wetland Restoration targets:
• 77% - 80% - 83% - 89% - 100%
- Prices (Tax imposed) of the proposed programs randomly
assigned
• $50 - $100 - $200 - $350 - $600
• Levels assessed in focus groups and pre-tests
Vote
The Current Trend
A Proposed Program
Results in further wetland loss: 77% of 1968
wetlands currently remain in southern
Manitoba, but this will decline to 70% (950,000
acres) by 2020.
2020
Maintain wetlands at their current level
through 2020, which is 77% (1,000,000 acres)
of 1968 levels in southern Manitoba
4500 semi-truck loads of fertilizer
5000 semi-truck loads of fertilizer
1.1 billion cubic meters of water
1.2 billion cubic meters of water
6 million tonnes of soil from being eroded
6.8 million tonnes of soil from being eroded
58,000 breeding pairs of ducks
63,000 breeding pairs of ducks
740,000 cars
800,000 cars
$0 annually for 5 years
$
Wetland Area Targets
Water Quality
Q
y
By 2020 wetlands will annually filter the
equivalent of about:
Flood Control
By 2020 wetlands will annually control about:
Soil Erosion
By 2020 wetlands will annually control about:
Wildlife Habitat
By 2020 wetlands will annually provide habitat
for about:
Carbon Capture and Storage
By 2020 wetlands will annually store carbon
equivalent to the emissions of about:
Your household’s annual share investment
paid through tax increases for the next 5
years, 2008-2012
annually for 5 years
The Referendum Question
Question 10a. Please carefully compare the two
alternatives presented in the table above. If you had to
vote
t on these
th
two
t
options,
ti
which
hi h one would
ld you choose?
h
?
Please treat independently from all other votes. Please
mark one box only.
Current Trend
Proposed Program
Question 10b. How confident are you that this is the
choice you would make if this was an actual
referendum? Choose one only.
y
–
–
–
–
1. Very uncertain
2. Somewhat uncertain
3. Somewhat certain
4. Very certain
Development of Survey
• Formed an Expert Advisory Group
– Wetland experts
experts, agriculture experts
– Used to assist in survey design (content)
• Held a focus groups of other experts
– “sniff-test” of content and process
• Held a focus group of graduate students
– Editing, formatting etc.
• Held 3 focus groups with randomly drawn
Manitobans (each group about 10-11 in size)
– 2 in Winnipeg, 1 in Brandon
• Held 1 focus group with agricultural producers
Development of Survey
• These procedures ensured:
– Consequentiality
C
ti lit off th
the voting
ti scenarios
i
– Construction of the vector of tax levels
– Ensured information was pertinent, balanced,
and not too onerous in length and depth
– Checked
Ch k d for
f bias
bi iin tterms off pushing
hi
respondents to vote YES for programs
– Time
Ti
ttaken
k to
t complete
l t survey
Survey Administration
• Mode of administration: Internet
– a number of different advantages
• Ipsos Reid contracted
– Maintains a panel of 10,000 Manitobans
– Refreshes panel to ensure representativeness
• Two pilot samples used to further test
– 353 and 446 respondents respectively
– Used to further adjust tax levels
• Full sample
p size - 2000 respondents
p
– 1000 in each version
General Results
• % of respondents who felt government
could do MORE to:
– 70%: protect the environment
– 77%: reduce crime
– 86%:
86% iimprove roads
d and
d hi
highways
h
– 86%: improve health care
– 66%: lower taxes
Familiarity with Environmental Issues
70%
60%
50%
40%
Wetlands in south
Climate change
g
Nutirent overload (LW)
30%
20%
10%
0%
Unfamiliar
Somewhat
unfamiliar
Somewhat
Familiar
Familiar
% respon
ndents un
ncertain
Levels of Voting Uncertainty Found
25
25
20
20
15
15
10
10
5
5
0
0
77
80
83
89
Restoration level
100
25
100 200 350 600
Tax level
Comparison with Other Studies
14
12
% votes
10
8
6
4
2
0
5
15 20 25 50 60 100 175 200 325 350 600
tax level
ONT
Wetlands
DFO
Comparison of YeaYea-Sayers
Per cantage
Yea‐Sayer
Yea
Sayer Comparison Between Valuation Studies
Comparison Between Valuation Studies
16
14
12
10
8
6
4
2
0
Wetland
Ontario
Study Which program attributes were
important in voting?
Extremelyy and Veryy Important
p
attributes:
Water Quality & Price of the proposed programs (taxes)
1800
1400
1200
1000
800
600
400
200
ta
xe
s
Ad
di
ti o
na
l
st
or
ag
e
C
Ha
bi
ta
t
W
i ld
l if
e
ro
sio
n
E
So
il
on
tro
l
C
Fl
oo
d
ua
l it
y
W
at
er
Q
E
xp
an
si
on
0
W
et
la
nd
No. of resp
pondents
1600
Econometric Analysis
• Develop explanators of voting behaviour
• Used a binary logit model which assesses
the probability of voting “YES”
YES
• Uses maximum likelihood procedures
• Can be fancy
– Random coefficients, latent class, panel data
procedures etc.
– But not right now
PB2
Per person welfare measures
Retention
Restore 80%
Restore 83%
Restore 89%
Restore 100%
All votes
pooled
First votes
only
All but first
votes
295.47
((17.78))
346.14
(17.94)
364.35
(18.33)
351 57
351.57
(18.08)
379.07
(18.41)
468.40
((72.25))
601.94
(84.36)
677.56
(87.67)
551 06
551.06
(79.54)
556.79
(78.95)
270.78
((17.80))
309.96
(17.94)
318.18
(17.95)
321 82
321.82
(18.33)
353.30
(18.56)
* 5000 draws from multivariate normal covariance matrix
Slide 33
PB2
These are the updated estimates I ran this morning using the same file I gave you for the "how others would vote" issues.
Peter Boxall, 1/11/2010
Basic Econometric Modeling
g
• Used just the “design” variables
– Restoration level, tax level
• Included demographic
g p
variables
– Income, male, age, household size
• Included possibly “endogenous”
endogenous variables
– membership in enviro orgs, visited a park in
Manitoba attitudinal score on New
Manitoba,
Environmental Paradigm Scale
Econometric Results
• Most results as expected!!
• Tax negative and highly significant
– People do not like to pay taxes ceterus
paribus
– Universally found in all economic studies
– Downward sloping demand curve
• Restoration p
positive and highly
g y significant
g
– Increased restoration is preferred
Results continued
• Income positive and significant
– Higher
g
incomes tend to vote for wetland p
programs
g
• Male negative and insignificant
g p
positive and significant
g
• Age
– Older people tend to vote for the programs
• Household size negative
g
and significant
g
– Larger households with more kids tend to vote against
programs
• Residency matters in some cases
– Brandon respondents tend to vote for wetland programs
– Rural
R l residence
id
and
d other
th ttowns nott significant
i ifi
t
Results continued
• Members of environmental orgs positive
and significant
– Members tend to vote for the wetland
programs
• Visiting a park in Manitoba positive and
significant
– Visitors tend to vote for the wetland programs
Final Valuation Estimates
U i llogarithmic
Using
ith i fform ffor wetland
tl d provision)
i i )
• Retention:
– $294.02/household/yr over 5 years
(SD=$11.56)
• Low Levels of Restoration:
– $315.67/household/yr
y over 5 yyears
– (SD= $8.33)
• High
g Levels of Restoration:
– $357.75/household/yr over 5 years
(SD=$15.57)
Gross Estimates
• These are per household estimates
• Need to multiply by number of
households in Manitoba as this is a
provincial survey
– 448,765 household; 2006 census
• Al
Also di
discounting
ti iissues b
because
payments over 5 years involved
5
t
WTPi
NPVi  
t
t 1 (1  r )
Household WTP x 5 years discounted x total # households
Retention
= $ 600 million
Low Restoration = $ 640 million
High Restoration = $ 730 million
Conclusions
• There seems to be significant demand for
wetland
tl d restoration
t ti
• Need to compare this with the cost and
supply-side information to develop
economically efficient programs
– NOTE the STC research to be reported
tomorrow is a start
The Research Team Thanks:
• ACAAF – Ducks Unlimited for funding support
and advice
• Deerwood Soil and Water Management
g
Association, and Bill Turner
• AAFC’s Watershed Evaluation of Beneficial
Management Practices Program
• Farm Level Policy Research Network
• Antony Samarawickrema, Orsolya Perger,
Marian Weber, Dana Harper and Alicia Entem