The effects of subjective perception on preference heterogeneity

ABSTRACT

Heterogeneity in preferences among individuals and groups for environmental goods has been widely observed in non-market valuation studies. However, the underlying mechanisms that produce the heterogeneity are not yet sufficiently understood, and the explanation remains one of the major questions in the stated preference economic analysis. This paper examines whether variations in perceived quality explain the observed heterogeneity in willingness-to-pay. Individuals’ willingness-to-pay for an environmental quality change is modelled as a function of their subjective perceived quality, instead of objective quality, and individual characteristics. At the same time, their subjective perception is modelled as a function of their individual characteristics. An empirical test is conducted using a choice experiment data set that evaluates endangered species conservation. The results indicate that (i) respondent’s subjective perception depends on their motivation, knowledge and comprehension; (ii) the willingness-to-pay estimate depends on the subjective perception as well as attitude, motivation and gender. The finding supports that respondent’s willingness-to-pay is dependent on their perceived quality, indicating that the subjective perception is one of the potential sources of the estimated preference heterogeneity. This suggests that prior knowledge as well as posterior understanding could have an effect on respondent’s willingness-to-pay through their subjective perception endogenously.

KEYWORDS:

• Preference heterogeneity
• perception formation
• endogenous bias
• choice experiment
• endangered species

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 This subjective evaluation (i.e. perception) of extinction risk was elicited on a 1 (high) to 10 (low) scale.

2 We use superscripts (p and w) for theoretical (conceptual) variables (i.e. z^p , z^w ) and subscripts for empirical variables (i.e. z_p, z_w).

3 In our application, respondent’s subjective perception of extinction risk was elicited on a 1 (high) to 10 (low) scale, and its functional form was specified by an ordered probit model.

4 The individual characteristic variables z consist of variables in z^p or variables in z^w , which can be decomposed into z^pw (variables in both z^p and z^w ), z^op (variables in z^p and not in z^w ) and z^ow (variables in z^w and not in z^p ): z = [z^pw z^op z^ow ]^T, z^p  = [z^pw z^op ]^T and z^w  = [z^pw z^ow ]^T. There is no assumption made about the relationship between z^p and z^w . By definitions, the followings are hold in the Taylor expansion: ∂e/∂z^w ≠ 0; ∂e/∂z^op = 0; ∂p/∂z^p ≠ 0; ∂p/∂z^ow = 0.

5 The panel consisted of residents of the Tokyo metropolitan area. Mitani et al. (2008, Table 2) summarized the respondent characteristics, with comparison to the population of the area.

6 We first used the choice experiment data set of Mitani, Shoji, and Kuriyama (2008) to estimate a Bayesian mixed logit model with six explanatory variables, including Cost (continuous) and Avoidable (dummy). We used 10,000 iterations for burn-in followed by additional 10,000 iterations after convergence had been reached, of which the draws in every 10th iterations were retained. Then, we derived individual-specific WTP values for each respondent.

7 These survey questions were designed to measure respondent’s psychometric characteristics. We use the principal component analysis to reduce the dimensions (i.e. the number of explanatory variables) and overcome the multicollinearity problem, which can be critical for the estimation of a simultaneous equation model.

8 We selected variables based on the convergence of the likelihood function and its log-likelihood values. To identify parameters, we also imposed exclusion restrictions in the WTP function as in a classical simultaneous equation model.

Additional information

Funding

This work was supported by Japan Society for the Promotion of Science: [Grant Number JSPS KAKENHI Grant Numbers 15J02390,JSPS KAKENHI Grant Numbers 18H00832].