Do divergences between stated and actual willingness to pay signify the existence of bias in contingent valuation surveys?
Introduction
Studies of the contingent valuation method (CVM) to determine consumers’ willingness to pay (WTP) for goods and services are increasingly being conducted to better understand the method and to aid decision-making. In sub-Saharan Africa and other developing regions, the emphasis has been on using CVM to understand demand and set prices for goods and services (Weaver et al., 1996; Whittington, Matsui-Santana, Freiberger, Van Houtven, & Pattanayak, 2002). In developed countries, the CVM has been predominantly used for generating consumers’ monetary values for goods and services for cost-benefit analysis (Klose, 1999; Birch, Gafni, & O’Brien, 1999). Some authors have found the CVM to be a valid method to determine WTP (Chestnut, Keller, Lambert, & Rowe, 1996; Onwujekwe, Chima, Shu, Nwagbo, & Okonkwo, 2001; Zillich, Blumenschein, Johannesson, & Freeman, 2002; Foreit & Foreit, 2003), while others have questioned the validity of the method (Mitchell & Carson, 1989; Diamond & Hausmann, 1994; Liljas & Blumenschein, 2000; Stavem, 2002).
The major criticism of the validity of CVM has been that stated WTP is a poor indicator of actual WTP (Diamond & Hausmann, 1994). The CVM is based on a hypothetical market in which respondents are not actually required to make the contributions they claim to be willing to pay (Foster, Bateman, & Harley, 1997). This has led to experiments to test the external validity of the method, by comparing stated and actual WTP and ascertaining whether the two diverge systematically (Foster et al., 1997). Most of the studies that have compared stated and actual WTP found divergences between the two (Corso, Hammitt, Graaham, Dieker, & Goldie, 2002; Blumenschein, Johannesson, & Blomquist, 1998; Blumenschein, Johannesson, Yokoyama, & Freeman, 2001; Onwujekwe, 2001; Johannesson, Liljas, & Johansson, 1998, 1999; Cummings, Harrison, & Rustrom (1996), Loomis, Brown, Lucero, & Peterson (1997); Cummings, Harrison, & Rustrom, 1995; Foster et al., 1997; Frykblom, 1997; Neill, Cummings, Ganderton, Harrison, & McGuckin, 1994).
Divergences in WTP could occur because of bias or because of legitimate changes in consumers’ values or circumstances. Distinguishing between the two is essential for better understanding the validity of CVM. At least six potential sources of bias have been identified (Foster et al., 1997; Dickie, Fisher, & Gerkins, 1987; Mitchell & Carson, 1989; Whittington, Briscoe, Mu, & Barron, 1990; Kealey et al., 1988). First, respondents might be influenced by the hypothetical nature of the situation, which might fail to provide sufficient incentive for respondents to reveal their true valuations. Second, the scenario could contain incentives for respondents to misrepresent their true WTP amount. For instance, if subjects perceive that the information could be used for price setting, they might understate their WTP. Third, the scenario may contain implied value cues that influence WTP amounts. Fourth, there may be mis-specification or mis-perception of the scenario, leading the respondent to understand or correctly perceive the characteristics of the good being described by the interviewer. Fifth, the individual may not take the contingent valuation questions seriously and simply respond by giving whatever answer occurs first to them. Finally, people may lack experience with aspects of CVM itself or lack knowledge about the good. Many, but not all, of these can be addressed through careful attention to experimental design.
Alternatively, there may be real changes in respondents’ valuations or circumstances, which affect their behaviour when the good in question is actually offered for sale. Whittington, Smith, Okoroafor, Liu, and McPhail (1992) studying WTP for water services in Nigeria debriefed respondents and found that they had logical reasons for changing their minds and concluded that there was no evidence of strategic bias in their study. Similarly, in determining test–retest reliability of WTP, Whitehead and Hoban (1999) found that there were socio-economic factors that genuinely accounted for the divergence between stated WTP at test and at retest.
In practice, calibration of stated WTP responses has been used to limit divergence between stated and actual WTP, by adjusting for the degree of certainty in a hypothetical yes response (Blumenschein, Johannesson, & Blomquist (1998), Blumenschein, Johannesson, Yokoyama and Freeman (2001)). For example, Johannesson, Liljas, & Johansson (1998), Johannesson et al. (1999) found that “definitely sure” yes responses were significantly lower than actual yes responses, and thus provided a lower bound for the estimate of real WTP. Similarly, Champs, Bishop, Brown, and McCollum (1997) found that hypothetical donations significantly exceeded real donations, but there were no significant differences if only subjects that were “very certain” of their yes response were counted as real yes responses.
One potential approach to improving the reliability and validity of CVM is increasing the realism of the contingent market in terms of specifying the item to be valued, describing the contingent market, explaining the method of payment and selecting a questioning format for eliciting values (Reiling, Boyle, Philips, & Anderson, 1990). Matthews, Rocchi, Wang, and Gafni (2001) posited that mimicking the live environment in CVM could help to improve validity of elicited WTP. Comparing across studies, Kealey et al. (1988) suggested that the discrepancy between hypothetical and actual CVM valuations is smaller when the commodity is more important to the respondent and that improving the realism of the CVM scenario could induce more intention-consistent or truthful behaviour.
This paper presents findings of stated and actual WTP for insecticide-treated bed-nets (ITNs) and explores the causes of divergences between the two. Specifically, it aims to identify whether divergences in WTP can be attributed to changes in socio-economic variables and other factors that influence WTP. Such evidence would support the view that not all divergence between stated and actual WTP is due to bias, and that changes in circumstances and values may be legitimate causes of such divergence. The divergences in WTP elicited with three question formats are examined. One of the question formats was structured haggling (SH), a novel Nigeria-specific technique designed as part of the study to increase the realism of the CVM question formats in the study area. It was hypothesised that a CVM question format that more closely resembled local patterns of bargaining for goods would more closely simulate actual purchasing thought processes and thereby reduce divergence between stated and actual WTP.
Section snippets
The sample
The study was conducted in three villages that have high malaria transmission rates all year namely; Amaetiti, Ahani and Enugu Akwu in Achi, Enugu State, Southeast Nigeria. The people are of “Igbo” ethnic group and Christianity is the major religion, with subsistence farming being their major source of livelihood. Bargaining for goods is the norm in local markets. Fixed prices operate in supermarkets and some shops although mainly in the cities and mostly for select people that live in the
General descriptive statistics
First survey: The usable numbers of questionnaires for the analysis were 261, 267 and 273 for the BG, BWFU and SH technique, respectively. In the three question formats, the majority of respondents were heads of households, middle aged, females, married and farmers. A minority of the respondents had formal education and the average years of formal education were 3.5, 4.0 and 5.1 in BG, BWFU and SH, respectively. A total of 75.1%, 80.5% and 75.5%, respectively, of the respondents in the BG, BWFU
Discussion
In contingent valuation, not all divergent WTP may be due to bias but because of justifiable determinants of demand change as was found in this study. Hence, divergences between stated and actual WTP should be regarded prima facie as an indicator of poor external (criterion) validity. Our results suggest that researchers need to determine why divergences in WTP exist, and to characterise the factors that could cause them. Divergences may be caused by legitimate factors that alter consumers’
Conclusion
There were genuine reasons people changed their minds about their WTP for ITNs, and thus not all WTP divergences could be said to be due to strategic or hypothetical bias. Therefore, “in order to better understand why hypothetical and real CV valuations may differ, it is important to analyse the motivation in individual answers to hypothetical questions and develop a theory of how individuals respond to hypothetical questions” (Blumenschein et al., 1998). These divergences would also inform
Acknowledgements
We are grateful to the Health Policy Unit, LSHTM for funding the writing-up of this paper. This study received financial support from the UNDP/World Bank/WHO Special Programme for Research and Training in Tropical diseases. Kara Hanson and Julia Fox-Rushby are members of the Health Economics and Financing Programme, which is funded by the UK Department for International Development.
References (36)
- et al.
The reliability and validity of willingness to pay surveys for reproductive health pricing decisions in developing countries
Health Policy
(2003) Hypothetical question modes and real willingness to pay
Journal of Environmental Economics and Management
(1997)The contingent valuation method in health care
Health Policy
(1999)- et al.
On hypothetical bias and calibration in cost-benefit studies
Health Policy
(2000) - et al.
Use of an interactive tool to assess patients’ willingness to pay
Journal of Biomedical Informatics
(2001) - et al.
Helicopters, hearts and hipsUsing willingness to pay to set priorities for public sector health care programmes
Social Science & Medicine
(1998) Criterion and content validity of a novel structured haggling contingent valuation question format versus the bidding game and binary with follow-up questions
Social Science & Medicine
(2004)- et al.
Investigating starting-point biasA survey of willingness to pay for insecticide-treated nets
Social Science & Medicine
(2002) - et al.
Willingness to pay for child survivalResults of a national survey in a Central African Republic
Social Science & Medicine
(1996) - et al.
Private demand for a HIV/AIDS vaccineEvidence from Guadalajara, Mexico
Vaccine
(2002)
Giving respondents time to think in contingent valuation studies
Journal of Environmental Economics and Management
Experimental results on expressed certainty and hypothetical bias in contingent valuation
Southern Economic Journal
Hypothetical versus real willingness to pay in the health care sectorResults from a field experiment
Journal of Health Economics
Willingness to pay and the valuation of programmes for the prevention and control of influenza
Pharmacoeconomics
Using donation mechanisms to value non-use benefits from public goods
Journal of Environmental Economics and Management
Measuring heart patients’ willingness to pay for changes in angina symptoms
Medical Decision Making
Assessing preferences for prevention versus treatment using willingness to pay
Medical Decision Making
Home-grown values and hypothetical surveysIs the dichotomous choice approach incentive-compatible?
American Economic Review
Cited by (29)
Does improved risk information increase the value of cholera prevention? An analysis of stated vaccine demand in slum areas of urban Bangladesh
2021, Social Science and MedicineCitation Excerpt :Nevertheless, CVM remains a dominant and important valuation method for goods and services with incomplete markets and through both preemptive careful study design and enumerator training and post-study evaluation of coping or averting behaviors, CVM-generated WTP estimates can provide valuable insight into demand for currently unavailable or inaccessible goods and services (Orgill-Meyer et al., 2018). For example, including context-specific scenarios (Shono et al., 2014) and comparing CVM estimates to actual cost estimates (Onwujekwe et al., 2005) can help to estimate the presence and size of bias in stated WTP estimates. While CVM offers the flexibility to estimate WTP for cholera vaccines across a large population, even if that population has limited or no access to them, revealed preference valuation methods utilize prior household resource allocations and decisions to inform the valuation estimate.
The criterion validity of willingness to pay methods: A systematic review and meta-analysis of the evidence
2019, Social Science and MedicineCitation Excerpt :The majority of comparisons (n = 94), did not explicitly use any specific terms for validity assessment, preferring to reflect papers as testing comparisons between hypothetical and actual WTP values. Approximately one fifth (n = 32) referred to this as testing for hypothetical bias (Blumenschein et al., 2014; Botelho and Pinto, 2002; Bryan and Jowett, 2010; Camacho-Cuena et al., 2004; Getzner, 2000; Johannesson, 1997; Mozumder and Berrens, 2007; Murphy Stevens et al., 2002; Onwujekwe et al., 2005). Two comparisons from the same study used the term predictive validity (Onwujekwe, 2001), while one used external validity (Muller and Ruffieux, 2011).
Willingness to pay for electric boats in a protected area in Italy: A sustainable tourism perspective
2019, Journal of Cleaner ProductionCitation Excerpt :as a % of the ticket) (Option 2) Uncertainty in the elicitation format has been introduced to overcome criticism of the validity of CV method for goods and services, according to which stated WTP might be a poor indicator of actual WTP due to bias or changes in consumers’ values or circumstances (Whitehead et al., 2008; Onwujekwe et al., 2005). A modified payment card method (Voltaire et al., 2013) has been used, thus allowing respondents to choose jointly WTP and degree of certainty (Fig. 2).
Comparing the validity of the payment card and structured haggling willingness to pay methods: The case of a diabetes prevention program in rural Kenya
2016, Social Science and MedicineCitation Excerpt :This is important because in the rural African marketplace, similar to a flea market transaction, once a price is agreed upon, it is considered unacceptable for the seller to haggle the price upwards. To date, SH has only been published in a limited number of contingent valuation studies by its developer from 2003 to 2008, where it has been directly compared to the BG in malaria treatment (Onwujekwe et al., 2004b) and with the BG and DC with follow up (DCFU) in malaria prevention (Onwujekwe, 2004; Onwujekwe et al., 2003, 2004a, 2005a, 2005b, 2008). In summary, these studies showed SH: had a mean bid lower than DCFU but approximately 10% greater than BG, and greater theoretical validity than BG and DCFU.
Voluntary counseling and testing (VCT) for HIV/AIDS: A study of the knowledge, awareness and willingness to pay for VCT among students in tertiary institutions in Enugu State Nigeria
2011, Health PolicyCitation Excerpt :In sub-Saharan Africa the CVM approach has been used most often to understand the demand and set prices for goods and services [36] unlike in high income countries where CVM is used mostly for generating the monetary values consumers place on goods and services for cost benefit analysis [37,38]. However, the validity of CVM has often been criticized because it is based on a hypothetical market in which respondents are not actually required to make the contributions they claim to be willing to pay because hypothetical donations have been found to significantly exceed real donations unless respondents that actually responded were very certain of their response and stuck by it [39–41]. In order to improve the reliability and validity of CVM, a more realistic scenario is created by specifying the item to be valued, describing the contingent market, explaining method of payment and selecting questionnaire formats for eliciting values [42].