Do preferences to reduce health risks related to air pollution depend on illness type? Evidence from a choice experiment in Beijing, China

https://doi.org/10.1016/j.jeem.2020.102355Get rights and content

Abstract

This study elicits preferences for clean air in a fast-developing context with increasing regulatory efforts and widespread adoption of self-protection measures against air pollution. We examine whether willingness to pay (WTP) to reduce mortality and morbidity risk depends on the type of illness caused by the pollution. Three major illnesses attributable to air pollution are examined in a choice experiment in Beijing, China. We find robust evidence, testing for both observed and unobserved preference heterogeneity, that WTP does not vary by illness type, and hence, that WTP for policy purposes should not be differentiated based on illness type. We also find that income, education, gender and other factors related with risk vulnerability well predict self-protection, and that respondents who engage more in self-protection have stronger preferences for public interventions. Our results suggest a value of a statistical life (VSL) and value of a statistical illness (VSI) of RMB 5.54 million (USD 1.58 million) and RMB 0.82 million (USD 0.23 million), which are higher than earlier estimates in China. This imply that for societies with strong economic growth and significant pollution, VSL and VSI are likely to increase rapidly, further strengthening the role of policies on pollution control and public health.

Introduction

Ambient air pollution is a major environmental hazard to human health (Dockery et al., 1993; Pope III et al., 2002; Lim et al., 2012; Landrigan et al., 2018). The Global Burden of Disease Study (2017) estimates that 7.2 million premature deaths in 2017 were due to environmental factors, of which 3.4 million were due to ambient fine particulate matter (PM2.5) and ozone pollution. Most of these environmentally caused illnesses and fatalities occur in highly populous emerging economies, where over the past few years managing air quality has become an increasingly significant regulatory concern (China State Council, 2013; India MEFCC, 2019). One example of a country undergoing this profound change is China, a very large economy with global impact. Since the passage of the Air Pollution Prevention and Control Action Plan (Action Plan) in 2013 and the greatly amended Law on the Prevention and Control of Atmospheric Pollution (New Law) in 2015, both China's central and local governments have implemented numerous large-scale, and often costly, policies designed to control air pollution.1 The Action Plan and the New Law are not the first measures taken to deal with air pollution in China, but they are the first to explicitly aim to reduce PM2.5 concentration through a combination of political incentives and more rigorous enforcement (Jin et al., 2016).

Such regulatory efforts come at a cost, however. One tool favored by economists for evaluating policies is benefit-cost analysis (BCA). As a decision tool, BCA provides decision makers with guidance for efficient resource allocation. The strong efficiency criterion for regulatory efforts to reduce air pollution is to maximize the net benefits (benefits minus costs). A weaker efficiency criterion requires that benefits “justify” costs, which has been embedded in legislation in many developed countries for decades (US OMB, 2003; Wiener, 2006; Morgenstern, 2014). In China, however, BCA, or more broadly, regulatory impact analysis (RIA), has seldom been used to evaluate environmental policies (Jin et al., 2016, 2017). Such an approach is urgently needed, as there is a high risk of resource misallocation when air pollution control policies are designed and implemented without timely evaluation of their efficiency. Lack of policy evaluation is not only a concern for China: due to its size, China's policies can have a large global impact, e.g. by affecting the prospects of climate change mitigation.

One challenge when implementing evaluation tools like BCA is that no easily available prices exist for health outcomes. Benefit-cost analysis needs a common metric for all benefits and costs, and usually money acts as this common metric. As no easily available prices exist, health outcomes need to be monetized by analysts. Monetary estimates for health benefits should be based on the individual preferences of the affected population, and the appropriate measure of such preferences is individual willingness to pay (WTP) for health risk reductions (see, e.g., Andersson et al., 2019a). Empirically, WTP is elicited either through revealed-preference (RP) or stated-preference (SP) studies, in which actual individual choices in markets or their answers to hypothetical scenarios in surveys are used (see, e.g., Freeman et al., 2014; Champ et al., 2017). Concerning WTP for health risk reductions, much of the focus has been on eliciting values to reduce mortality risk. The monetary value is conventionally reported for the prevention of one premature statistical death in the population, which is usually referred to as value of a statistical life (VSL) (Andersson et al., 2019a).2 However, many health outcomes, including those related to air pollution, are non-fatal. Such non-fatal outcomes can also be monetized using the same RP and SP methods (see, e.g., Barwick et al., 2018; Hammitt and Haninger, 2017), and their monetary equivalents to the VSL are often referred to as value of a statistical illness (VSI) (see, e.g., Andersson et al., 2015).

The aim of this article is to elicit individual preferences regarding health risk reductions resulting from air quality improvement in a fast-developing metropolitan context. The study is conducted in China, and one of its three main objectives is to elicit VSL and VSI that can inform policy development in a Chinese context. There is already a rich literature on monetizing health preferences (see, e.g., Viscusi, 2014; Lindhjem et al., 2011), but the majority of the studies have been conducted in developed countries, and it is well established that there is no one-size-fits-all “best estimate” of, e.g., VSL (Cameron, 2010). The environmental problems in developed countries are often less severe than in developing ones, and therefore the associated health risks are lower. Other factors such as income, age, gender, and education level may also influence individual preferences concerning health, and have been examined in various studies (see, e.g., Carlsson et al., 2010; Alberini and Ščasnỳ, 2011; Cameron and DeShazo, 2013). Given that 92% of all pollution-related mortalities occur in low- and middle-income countries (Landrigan et al., 2018), and that various factors influence VSL, using value transfers based on preference estimates from developed to developing countries (Masterman and Viscusi, 2018) involve substantial uncertainty (Robinson et al., 2019b). Moreover, unlike mortality, WTP for morbidity risks reductions is poorly understood in developing and developed countries (US EPA, 2010; Cameron, 2014; Hunt et al., 2016). Further, it can be argued that existing empirical findings on WTP to reduce health risk related to air pollution in China are insufficient to support the current regulatory context. Huang et al. (2018), in addition to providing new estimates, provided a review of Chinese VSL studies, which revealed that many studies were relatively old. However, due to China's rapid development, pollution and income levels have changed; in addition, through different information channels people have become more aware of air pollution and its effects on health (Yang, 2016; Ravetti et al., 2019). For these reasons, preferences regarding better air quality and health risk reductions may have changed in China, and hence people's WTP (Hammitt et al., 2019).

The second main objective of this study is to examine whether the type of illness matters for individual preferences, i.e. whether their WTP to reduce the mortality or morbidity risk differs depending on illness type. This is of high research and policy relevance, because if preferences differ it would suggest that VSL and VSI should be differentiated based on illness type. To conduct the empirical analysis and elicit WTP we employ choice experiments (CE). This is an SP method; we choose to use this approach rather than RP because it allows us to control the decision alternatives. In addition, RP studies on morbidity are difficult to conduct, because real market data rarely reflect the trade-offs between often non-fatal, chronic environmental morbidity and wealth. An alternative SP approach is the contingent valuation (CV) method, which values the aggregate outcome of an environmental change; it is therefore less suited to isolating separate values for the mortality and morbidity components (Adamowicz et al., 2011). To test if preferences differ across illnesses, it should be noted that our study is from an environmental health-risk “prevention” perspective. Related but differently, there are also CE studies in health economics eliciting preferences across illness profiles (e.g., symptoms, treatments and lifetime impacts). Medical and health-care contexts motivate them to use such within-illness characteristics as attributes; see, e.g., the discussion of this literature in Cameron (2014). We instead used illness type as part of scenarios. This is because under environmental pollution context, better environment mainly leads to aggregated reduction of illness specific risks, rather than changes of within-illness characteristics. Hence, given our objectives, we designed a CE to examine individual preferences for mortality and morbidity risk reductions of different illnesses from publicly funded measures to improve air quality in a rapid developing, severely polluted, strong regulatory context, i.e. Beijing, China.

Our third objective is to examine the relationship between self-protection and preferences for publicly provided policies.3 In 2013 the real time air quality index became available, and since then air pollution and its potential negative health effects have been widely known to Chinese citizens (Jin et al., 2016). This has lead to a widespread adoption of self-protection measures, such as wearing particulate-filtering masks (Zhang and Mu, 2018) and using household air-purifiers (Ito and Zhang, 2020). Different individuals, including our respondents, adopt these measures with different type of products, and at different prices and usage frequencies (Sun et al., 2017). These directly observed behavioral indicators jointly point to a wide distribution of a latent self-protection “construct” (or attitude/trait) across respondents. In this study, we estimate such latent self-protection and use it to explain individual choices of public programs. Our contextual analysis informs resource allocation among societal groups with different self-protection levels against air pollution.

This article is structured as follows. In the next section we first describe health risks related to air pollution that are of interest to this study, then the theory behind monetizing health risk reductions, and finally the effect of self-protection on the WTP. In section 3, we describe the survey used to collect the data and how it was implemented. In sections 4 Empirical methods, 5 Results, we describe the empirical methods used and the results of the analysis. The article ends with section 6, which discusses the findings and their implications.

Section snippets

Air pollution and its effects on health risks

Our main objectives are to elicit WTP for both non-fatal morbidity (VSI) and mortality risk reductions (VSL) from ambient air quality improvement, and to examine whether the values differ across illnesses. Among the various adverse health effects associated with ambient PM2.5 and ozone pollution, the ones with strongest epidemiological evidence are four major adult illnesses, ischemic heart disease (IHD), cerebrovascular disease (stroke), chronic obstructive pulmonary disease (COPD) and lung

Survey structure and administration

We designed a survey to elicit individual WTP for the public provision of health risk reductions through air quality improvement. Data were collected by the survey company WJX during fall 2016 using a Web survey. Respondents were recruited from a panel consisting of more than two million randomly recruited Internet users all over China. We restricted our respondents to residents of Beijing, which has the highest Internet penetration rate in China, to mitigate the risk of a non-representative

Utility function and WTP

Our econometric analysis starts from a random utility model with utility Unjt for respondent n to choose alternative j from J = 3 alternatives (two programs and the status quo) in choice situation t from a total of T = 8 choice sets defined as,Unjt=Vnjt+ɛnjt,with the unobserved part of utility ɛnjt considered to be an IID type I extreme value, and the observed part of utility (Vnjt) defined as,Vnjt=asc+β1morbnjtexp(β3delaynjt)+β2mortnjtexp(β3delaynjt)+β4costnjt,where asc is an

Descriptive statistics

Table 3 provides the descriptive statistics for selected respondent characteristics. Our sample is Internet-enabled, and is younger and more highly educated than the average population aged 20 and over in Beijing. It is common in surveys to find that more highly educated are over-represented, which in this context may explain also the larger share of young respondents compared with the general population. Therefore, if these characteristics are correlated with the respondents’ preferences, the

Discussion

In this study, we use CE to elicit preference for health risk reduction through air quality improvement in a fast-developing metropolitan context with high pollution levels and therefore a high background risk, i.e. Beijing, China. The affected population is increasingly wealthy, more aware of environmental health problems, and there is widespread adoption of self-protection measures. The first objective of this study was to provide robust WTP measures for mortality and morbidity

Declaration of competing interest

The authors declare that there is no conflict of interest.

Acknowledgements

We thank Wiktor Adamowicz, Anna Alberini, the participants of the Ulvön 2018 conference, WCERE 2018, SBCA 2018 conference, two anonymous reviewers and the editor for helpful comments on prior versions of this paper. Yana Jin acknowledges the financial support from Environmental Science and Policy Mellon Postdoctoral Fellowship at William & Mary, Global Research Institute at William & Mary, the Mälor Scholarship at the Beijer Institute of Ecological Economics, and China Scholarship Council.

References (86)

  • S. Hoffmann et al.

    The willingness to pay for mortality risk reductions in Mongolia

    Resour. Energy Econ.

    (2012)
  • A.R. Hole

    Modelling heterogeneity in patients' preferences for the attributes of a general practitioner appointment

    J. Health Econ.

    (2008)
  • N. Jacquemet et al.

    Preference elicitation under oath

    J. Environ. Econ. Manag.

    (2013)
  • Y. Konishi et al.

    A framework for estimating willingness-to-pay to avoid endogenous environmental risks

    Resour. Energy Econ.

    (2011)
  • P.J. Landrigan et al.

    The Lancet Commission on pollution and health

    Lancet

    (2018)
  • A.M. Leiter et al.

    Risky sports and the value of safety information

    J. Econ. Behav. Organ.

    (2016)
  • S.S. Lim et al.

    A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010

    Lancet

    (2012)
  • L. Liu et al.

    Endogenous private safety investment and the willingness to pay for mortality risk reductions

    Eur. Econ. Rev.

    (2006)
  • J. Meyerhoff et al.

    Learning how to choose—effects of instructional choice sets in discrete choice experiments

    Resour. Energy Econ.

    (2015)
  • J. Orgill-Meyer et al.

    Comparing contingent valuation and averting expenditure estimates of the costs of irregular water supply

    Ecol. Econ.

    (2018)
  • C. Sun et al.

    Self-protection investment exacerbates air pollution exposure inequality in urban China

    Ecol. Econ.

    (2017)
  • A. Vij et al.

    How, when and why integrated choice and latent variable models are latently useful

    Transp. Res. Part B Methodol.

    (2016)
  • J. Walker et al.

    Generalized random utility model

    Math. Soc. Sci.

    (2002)
  • J. Zhang et al.

    Air pollution and defensive expenditures: evidence from particulate-filtering facemasks

    J. Environ. Econ. Manag.

    (2018)
  • A. Alberini et al.

    Context and the VSL: evidence from a stated preference study in Italy and the Czech Republic

    Environ. Resour. Econ.

    (2011)
  • H. Andersson et al.

    Willingness to pay and QALYs: what can we learn about valuing foodborne risk?

    J. Agric. Econ.

    (2015)
  • H. Andersson et al.

    Valuation of health risk. Oxford Research Encyclopedia of Economics and Finance

    (2019)
  • F. Bahamonde-Birke et al.

    Is sequential estimation a suitable second best for estimation of hybrid choice models?

    Transport. Res. Rec.

    (2014)
  • F.J. Bahamonde-Birke et al.

    About attitudes and perceptions: finding the proper way to consider latent variables in discrete choice models

    Transportation

    (2017)
  • P.J. Barwick et al.

    The Morbidity Cost of Air Pollution: Evidence from Consumer Spending in China

    (2018)
  • Beijing Health Bureau

    White Paper on Health in Beijing

    (2012)
  • M. Bierlaire

    Estimating hybrid choice models with the new version of biogeme

  • M. Bierlaire

    Estimating Choice Models with Latent Variables with Pandasbiogeme

    (2018)
  • B. BMBS

    Beijing Statistical Yearbook

    (2017)
  • T.A. Cameron

    Euthanizing the value of a statistical life

    Rev. Environ. Econ. Pol.

    (2010)
  • T.A. Cameron

    Valuing morbidity in environmental benefit-cost analysis

    Annu. Rev. Resour. Econ.

    (2014)
  • T.A. Cameron et al.

    Supplementary Materials to Accompany for “Demand for Health Risk Reductions”

    (2010)
  • D. Campbell et al.

    Non-attendance to attributes in environmental choice analysis: a latent class specification

    J. Environ. Plann. Manag.

    (2011)
  • F. Carlsson et al.

    Value of statistical life and cause of accident: a choice experiment

    Risk Anal.: Int. J.

    (2010)
  • P. Champ et al.

    A Primer on Nonmarket Valuation

    (2017)
  • China State Council

    Air Pollution Prevention and Control Action Plan

    (2013)
  • P.S. Corso et al.

    Valuing mortality-risk reduction: using visual aids to improve the validity of contingent valuation

    J. Risk Uncertain.

    (2001)
  • R.G. Cummings et al.

    Unbiased value estimates for environmental goods: a cheap talk design for the contingent valuation method

    Am. Econ. Rev.

    (1999)
  • Cited by (47)

    View all citing articles on Scopus
    View full text