E-Locus - Institutional Repository of the University of Crete

Home Collections School/Department School of Social Sciences Department of Economics Doctoral theses

Doctoral theses

Current Record: 2 of 31

[Add to Basket]

Identifier

000455140

Title

Overcoming hypothetical bias in contingent valuation surveys / Yakinthi G. Pavlaki.

Alternative Title

Διορθώνοντας το πρόβλημα της υποθετικής μεροληψίας σε έρευνες πιθανολογικής αποτίμησης

Author

Παυλάκη, Υακίνθη

Thesis advisor

Τζίνιους Μαργαρίτα

Reviewer

Τζουβελέκας Βαγγέλης
Τσιώτας Γεώργιος

Abstract

This Ph.D. thesis aims to build econometric models that can overcome hypothetical bias in Contingent Valuation surveys. Within 4 interrelated Chapters, this thesis focuses on constructing a mixture model and applying stochastic frontier analysis in order to include the existence of hypothetical bias. The main idea of the proposed model of the present thesis is that in a Contingent Valuation survey there might be two types of respondents. The first one refers to respondents that answer sincerely about their WTP and the second to respondents that overstate their WTP. The first chapter presents a literature review regarding Contingent Valuation Method (CVM) and the problem of Hypothetical Bias. Additionally, it analyzes the theoretical framework regarding the statistical models that are constructed in Chapter 2 and 3. The second and third Chapters present the proposed model that can overcome hypothetical bias for the open-ended elicitation format and for the double-bounded dichotomous choice elicitation format. The Chapters contain the theoretical background of the corresponding format, the stochastic frontier theory, which is applied to model hypothetical bias, the mixture models theory that is used in order to allow the existence of the 2 classes and the estimation procedure, including the EM algorithm. In order to test the validity of the proposed model, simulations took place for 1000 observations and 1000 replications for several cases. Additionally in Chapter 3, after the illustration of the simulation results an empirical application is presented with real CVM data. The fourth Chapter deals with the issue of the selection of starting values for the EM algorithm. Chapter 4 begins with the literature review about the initialization techniques and the importance of the initial values. Furthermore, application of three initialization methods were applied to the proposed model for the double-bounded format and a comparison of the initialization techniques took place in order to conclude which initialization strategy performs better.

Language

English

Subject

Composed Error

Contingent Valuation Method (CVM)

Double-Bounded

EM algorithm

Environmental Econometrics

Finite Mixture Models

Hypothetical Bias

Initial Values

Open-Ended