Priming and the value of a statistical life: A cross country comparison
Introduction
Air pollution not only causes negative environmental effects but is also responsible for several negative health effects. It has been shown to cause, e.g., mental health issues (Lu, 2020, Zhang et al., 2017) and respiratory diseases (Beatty and Shimshack, 2014, Ruchiraset and Tantrakarnapa, 2020), but that it also can lead to premature death (Lelieveld, Evans, Fnais, Giannadaki, & Pozzer, 2015). According to some recent studies, the number of annual premature deaths due to fine particulate ambient air pollution are between 3.4 and 8.9 million (Burnett et al., 2018, Global Burden of Disease Study, 2017).
Improving air quality therefore has the potential to provide large social benefits from reducing health risks related to air pollution. Such policies come at a cost, though, and to ensure that society’s resources are used efficiently the policies should be evaluated. One economic tool that can be used to provide information on whether the benefits of the measures exceed the costs is cost–benefit analysis (CBA). This tool favored by economists is founded in welfare theory and if the analysis shows that the benefits are larger than the costs, then it suggests that the policy is desirable for society. However, the use of CBA requires that all benefits and costs are measured in a common metric, which is usually money. Hence, it is necessary to monetize the different effects from the policy that do not have easily observable prices, like the reduced risk of dying from air pollution.
To empirically elicit preferences for health risk reductions and converting them into monetary values analysts rely on revealed- (RP) or stated-preference (SP) methods (see, e.g., Freeman, Herriges, & Kling, 2014). Whereas the former, i.e. RP methods, rely on individuals’ actual decisions in markets, such as accepting to take a riskier job if financially compensated (Gentry & Viscusi, 2016), or paying more for a property in an area with better air quality (Chay & Greenstone, 2005), the latter, i.e. SP methods, rely on individuals’ answers from hypothetical scenarios to measure their preferences. Economists have traditionally favored the RP approach since it relies on actual decisions. However, weaknesses with the approach include the necessity of markets (which can be an issue with public goods), access to good data (e.g. different environmental effects like noise and air pollution may be highly correlated making it difficult to identify the preferences for one of the effects), and the assumption that individuals make well-informed decisions in markets. Therefore, SP methods have gained ground due to their flexibility which allows them to construct the market scenario of interest, and to control the choice situations providing transparency for the analyst of the respondents’ decision alternatives. The hypothetical nature of the SP approach is its main weakness, though, but how to address this has been a major research area and several manuals and guidelines on best practice for SP methods are available, including the recent one by Johnston et al. (2017).
Among SP methods, the two techniques that dominate are contingent valuation (CV) and discrete choice experiments (DCE) (see, e.g., Champ, Boyle, & Brown, 2017). Mahieu, Andersson, Beaumais, Crastes dit Sourd, Hess, and Wolff (2017) showed in their review that DCE have gained popularity among economists to elicit respondents’ willingness to pay (WTP) for different non-market goods. The use of DCE has also gained ground when it comes to monetizing the value for health risk reductions (e.g., Adamowicz et al., 2011, Andersson et al., 2016, Cameron et al., 2010, Jin et al., 2020). In many of the studies estimating the WTP for health risk reductions the focus has been on eliciting WTP to reduce mortality risks, which is commonly normalized and referred to as the value of a statistical life (VSL) (Andersson, Hole, & Svensson, 2019). In this study, in addition to eliciting the WTP for a policy that aims at reducing the mortality risk due to ambient air pollution, we use priming treatments (e.g., Carlsson et al., 2013, Jacquemet et al., 2013) to assess whether they may influence the respondents’ answers when choosing between different policies in DCE. We focus on this type of mechanism as it is simple to implement and, if effective, can be of particular interest for practitioners to obtain reliable results. Our hypotheses in this paper are that priming, depending on the type of priming, can either address the issue of hypothetical bias in SP studies, and hence provide a more conservative (lower) WTP for the good of interest in the survey, or push respondents to care more for the good of interest, and hence, result in a higher WTP.
The priming treatments are described in more detail later in the paper but we focus on two types: (i) an oath presented at the beginning of the survey, and (ii) a priming scenario that combines information and questions on the respondents’ experiences. Regarding the potential difference in how the priming may influence among different populations we implement the same survey in the United States (US) and the United Kingdom (UK). We focus on these two countries for two reasons: (i) they share many common features, including the language, which mitigate the risk of too many confounding factors influencing the results, and (ii) the US are recognized as having a “distinctive culture” (Sunstein & Reisch, 2019) that may suggest that findings from that country may not be generalized even to a country that may be perceived as “similar” as the UK. Indeed, except in Carlsson et al. (2013) where the authors compare the effect of an oath in a survey in China and Sweden, evidence is lacking regarding the effect of such priming mechanisms to induce reliable results in SP studies.
The contributions of our study are: (1) since priming may influence respondents’ answers in SP studies, and hence estimated values to be used for policy purposes, it is of importance to examine whether priming does have an effect, and how robust these effects are between contexts (e.g. countries); and (2) in addition to what have already been tested in the literature, we extend the analyses by comparing what has already been used together with different forms of priming. Hence, the aim of the study is to be of both policy and research relevance by contributing to the literature on the behavioral aspects related to the design of SP studies that can have an effect on elicited preference estimates, such as the VSL, through the comparison of the effects of different forms of priming.
In the following section we briefly describe the concept of valuing mortality risk reductions and behavioral aspects of relevance when eliciting preferences in SP studies. After that, in Section 3, we describe the survey, i.e., how data were collected and the structure and content of it. Section 4 contains the model and empirical methods used followed by the results section. The paper ends with a discussion and some concluding comments.
Section snippets
Valuing safety and behavioral influences
In this section we first briefly describe the theory behind the VSL, the monetary value of interest to this study. We then discuss different behavioral aspects of relevance to SP studies that may influence respondents when they make their choices, and hence may affect the estimated monetary values.
Survey
The survey was programmed in LimeSurvey and implemented using MTurk in November and December of 2019 in the US and UK. In the past, there were some concern regarding the use of Mturk for economic experiments. However, several studies have replicated findings from lab experiments on MTurk (e.g., Arechar et al., 2018, Coppock et al., 2018, Gandullia et al., 2020) and, nowadays, this platform is regularly used to conduct online experiments (e.g., Gandullia, 2019, Horton et al., 2011, Jacquemet et
Baseline model
The respondents are assumed to derive utility from the attributes of the choice situations and their choices are analyzed using the Random Utility Model (RUM) (McFadden, 1974). According to the RUM, respondents maximize their utility, but the true one being unknown, a random component is considered. As previously explained, respondents were asked to choose their preferred option among alternatives (two policy options and one status quo) in choice sets. Individual ’s utility for
Data
In total, our sample consists of 806 US respondents and 694 UK respondents, corresponding to 1500 respondents.6 As described in Appendix A, we were able to identify respondents who gave inconsistent answers during the training session which we decided to use to remove those respondents who did from
Discussion
In this study, we use a DCE to elicit US and UK respondents’ preferences for reducing mortality risks related to air pollution. In addition, we test whether the respondents’ preferences, when choosing between different policies, can be influenced using two types of priming — two versions of an oath presented at the beginning of the survey, and a priming scenario that combines information and questions on the respondents’ experiences.
We used self-recruited MTurkers to conduct the study. A
Conclusions
Our results highlight a heterogeneity in respondents’ reaction to different priming. In a sense, we can make a link with the literature that focuses on the assessment of the effects of nudges and, in particular, on the effect of social norm comparisons (Allcott, 2011, Brent et al., 2015, Ferraro and Price, 2013).11 Similarly in SP studies, Ouvrard, Abildtrup and Stenger (2020) show
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
We are grateful for valuable feedback from anonymous reviewers and participants during seminars and at conferences. Henrik Andersson acknowledges funding from ANR, France under grant ANR-17-EURE-0010 (Investissements d’Avenir program).
References (80)
- et al.
Cheap talk reconsidered: New evidence from cvm
Journal of Economic Behaviour and Organization
(2006) - et al.
Valuation of cancer and microbial disease risk reductions in municipal drinking water: An analysis of risk context using multiple valuation methods
Journal of Environmental Economics and Management
(2011) Social norms and energy conservation
Journal of Public Economics
(2011)- et al.
Valuation of small and multiple health risks: A critical analysis of sp data applied to food and water safety
Journal of Environmental Economics and Management
(2016) - et al.
Private and public willingness to pay for safety: A validity test
Accident Analysis and Prevention
(2019) - et al.
Air pollution and children’s respiratory health: A cohort analysis
Journal of Environmental Economics and Management
(2014) - et al.
Using cheap talk as a test of validity in choice experiments
Economics Letters
(2005) - et al.
The truth, the whole truth, and nothing but the truth—a multiple country test of an oath script
Journal of Economic Behaviour and Organization
(2013) - et al.
The effect of the solemn oath script in hypothetical choice experiment survey: A pilot study
Economics Letters
(2014) - et al.
Hypothetical bias in stated choice experiments: Is it a problem? And if so, how do we deal with it?
Transportation Research Part A: Policy and Practice
(2014)