ABSTRACT
European Carp is a species of freshwater fish that live in Australia's freshwater rivers, wetlands and lakes. It is a dominant pest species in Australia. As part of investigations into the potential release of the Cyprinid Herpes Virus 3 to control Carp numbers, a Choice Modelling (CM) study was undertaken to value the potential non-market environmental benefits of such a program. These potential benefits relate to increases in native fish populations, the area of healthy wetlands and waterbird populations. The CM study explicitly included a probability attribute to reflect the risk that various release options will deliver the identified environmental outcomes. Consequently, the willingness to pay (WTP) estimates derived from the choice models are expressed in terms of the expected values of outcomes: the physical outcomes multiplied by their probabilities of occurring. The preferred model of choice was a Random Parameter Logit model with a non-linear (cubed) transformation of the levels for each probability weighted environmental attribute. Consequently, estimates of WTP derived from the model increase as the level of an expected attribute increases. This non-linear form is consistent with respondents treating risk in a nonlinear fashion.
Acknowledgements
The study was funded by the Freshwater Research and Development Corporation as part of the National Carp Control Plan.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Potential negative impacts of the Carp control program were considered as part of the broader cost benefit analysis of the program (Hardaker et al. Citation2020).
2 The study is focused on the non-use values held by Australian households. The distribution of the Australian population is such that a random sample of the size sought would inevitably include only a very small number of regional/rural respondents. Hence conducting focus groups in a major urban centre is both relevant and pragmatic.
3 The NGene software programme was used to develop the experimental designs.
4 While the expected utility framework was used as the foundation for this study, many other modelling approaches were also explored. Following Rolfe and Windle (Citation2015), alternative models estimated included main environmental effects without interactions terms between attributes and probabilities and main effects models together with interactions terms between attributes and probability. Combined linear and nonlinear models were also considered. In the main effects RPL models estimated, the wetland and bird attributes parameters had the incorrect sign and were not significant. The combined main effects and interaction models also had insignificant wetland and bird attribute parameters as well as insignificant wetland and probability interactions and bird and probability interactions parameters. In the combined linear and nonlinear models estimated, the linear and nonlinear forms of attributes parameters all had different signs resulting in U shaped or inverted U shaped WTP for the environmental attributes. These alternative results were inconsistent with theory and led to the rejection of the alternative approaches.
5 It is important to note that the coefficients estimated for each of the attributes cannot be interpreted separately, apart from the importance of their signs, because they are each confounded with a scale parameter. The division of parameter estimates, as occurs in the estimation of the implicit prices, overcomes this issue because it involves the cancellation of the scale parameter that is included in all coefficient estimates.