A BoxCox doublehurdle model of wildlife valuation: the citizens perspective

1
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• Roberto Martínez-Espiñeira
• Department of Economics
• St. Francis Xavier University
• June 2004

2
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• The study
• Citizens versus consumers preferences in
  contingent valuation
• Valuing goods, valuing bads, valuing both
• The survey coyotes in Prince Edward Island
• Results
• Discussion

3
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• Nuisance wildlife can be controlled or
  removed
• Nuisance wildlife could also be saved by
  compensating affected parties
• How much are individuals willing to pay to
  conserve nuisance wildlife?
• Does everyone consider nuisance wildlife a good?

4
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• When responding to CVM questions, individuals are
  assumed adopt a strictly private viewpoint
• individuals responses are interpreted as
  expressions of consumer preferences

5
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• But respondents to CVM surveys could be stating
  their preferences as citizens based on some
  notion of social responsibility, rather than as
  consumers concerned with the maximization of
  individual welfare

6
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• CVM surveys about environmental preservation are
  likely to be dominated by citizen judgments
  concerning social goals and responsibilities,
  rather than by consumer preferences (Blamey et
  al. 1995, Sagoff, 1988)
• especially the context of wildlife valuation

7
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• If some (or all) respondents adopt the citizens
  perspective when participating in a CVM survey,
  the results cannot be considered compatible with
  theoretical measures of welfare such as WTP and
  consumer surplus

8
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• In this survey, respondents were asked how much
  they thought it would be reasonable to ask all
  residents to contribute to a program to protect
  coyotes based on compensating farmers for
  livestock losses through an annual tax
• So they all were explicitly prompted to adopt a
  citizens perspective

9
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• Respondents were prompted then to engage in a
  role as Homo Politicus with shared
  responsibility as modeled by Nyborg (2000)
• They were NOT answering about their conventional
  WTP

10
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• The theoretical aspects of the analysis and the
  policy implications of the results differ
  substantially from those of conventional CVM
• But individuals responses could still be modeled
  in a way similar to the way in which WTP
  responses are modeled

11
A Box-Cox double-hurdle model of wildlife
valuation the citizens perspective

• Of special interest is the distinction between
• factors that increase the probability that an
  individuals states a positive support coyote
  conservation and
• factors that affect the stated degree of support

12
The Survey

• A phone survey in Prince Edward Island (PEI)
• Where coyotes have made themselves comfortable
  during the last decades after having traveled on
  the frozen sea from Nova Scotia and Quebec
• They thrive in the rural farmland of the Island,
  facing no natural predators
• But they cause problems for farmers, particularly
  sheep breeders
• there are no deer, moose, or bears in PEI, before
  coyotes came, foxes would be the biggest predator
  around

13
The Survey

• Without prevention, sheep losses can become so
  severe that a sheep farm could become financially
  unviable
• Throughout North America millions of dollars have
  been spent attempting to eliminate coyotes and
  all efforts have been failures
• In PEI, there is still controversy between those
  who want the government to fund coyote
  extermination and or subsidize measures to avoid
  predation and those who think they do not pose a
  problem

14
(No Transcript)
15
(No Transcript)
16
Methodology

• The responses to the open-ended question included
  a good number of zeros for the tax response
• Usually this problem is dealt with by using Tobit
  analysis
• But the Tobit model only allows one type of zero
  observation a corner solution

17
Methodology

• This might be too restrictive
• some individuals would never state a positive
  amount (they consider, for example, that coyotes
  are a bad)
• Tobit model can be made more flexible by
  considering these non-supporters of conservation

18
Methodology

• I use a series of models that increase the
  flexibility of the Tobit
• I use likelihood-ratio tests to check that the
  most flexible one proves to be the best to
  explain the responses

19
Methodology

• The Double-Hurdle model allows for the
  possibility that the decisions of whether to
  state support for conservation and the degree of
  support affected by a different set of variables
• I use a Box-Cox transformation of the dependent
  variable to account for the its non-normal
  distribution
• And I correct for heteroskedasticity
• I also checked that the hypothesis of
  independence between the error terms in both
  equations could not be rejected

20
Degree of support, Given that support is more
than zero
Dichotomous Choice 1supportgt0
21
Results

• Effect of age
• respondents around the age of 39 would be the
  least likely to state a positive amount for
  coyote protection through tax
• The age variable affects the choice of tax
  differently with a negative sign on agesq

22
Results

• Effect of education
• For the level of tax the effect of education
  exhibits an inverted-U shape as education levels
  approach college degrees, stated tax diminishes
• In the participation model, a U-shape suggests
  that the negative effect of educat on
  participation bottoms out for a level of educat
  2. 285, slightly higher than the level at which
  the expected size of tax is maximum (educat 1.866)

23
Results

• Effect of income
• Presents a negative sign for the level of tax and
  is non-significant in the participation equation
• Respondents were asked about their judgment on
  how much everyone should contribute to
  conservation. The answer has more to do with the
  expression of a political preference than with
  the combination of a preference and an individual
  ability to pay

24
Results

• Density presents a negative non-significant
• hunters stated support for significantly lower
  levels of tax to help protect coyotes
• However, hunters are much less likely to state a
  zero amount for tax

25
Results

• There is a significantly positive effect of the
  incometown variable those in richer counties are
  richer, tend to live in more urbanized areas, and
  probably have more favorable attitudes towards
  coyotes
• Effect of sheep ownership is also highly
  significant and positive. Farmers are usually
  disappointed with lethal methods for dealing with
  coyote predation and would of course be wiling to
  support high levels of general taxation earmarked
  to compensate them for losses to predation

26
Results

• There is a significantly positive effect of the
  incometown variable those in richer counties are
  richer, tend to live in more urbanized areas, and
  probably have more favorable attitudes towards
  coyotes
• Effect of sheep ownership is also highly
  significant and positive. Farmers are usually
  disappointed with lethal methods for dealing with
  coyote predation and would of course be wiling to
  support high levels of general taxation earmarked
  to compensate them for losses to predation

27
Conclusions

• some PEI residents would be willing to support an
  increase in general taxes to protect coyotes
• the effect of most variables on the stated
  socially acceptable increase in taxes goes in the
  same direction as the one they would be expected
  to have on the conventional willingness to pay
• One unsurprising exception is the case of
  household income

28
Conclusions

• The results stress the importance of modelling
  separately the decision on whether to support an
  environmental cause and the decision about how
  much to contribute to that cause.
• This is particularly important when considering
  species that some consider public bads rather
  than public goods.
• Many of the zero responses obtained are not
  simple corner solutions, but rather reflections
  of respondents who are not willing to support the
  protection of coyotes regardless of circumstances
  that might affect the degree of support of those
  who in principle support coyote protection

29
Conclusions

• The results should not be confused with those
  coming from a conventional contingent valuation
  study
• They are about what individuals perceive the
  government should do about the coyotes and should
  be used as a complement to other valuation
  exercises, not as a substitute.