1. Introduction
1.1 The Role of Contingent Valuation (CV) Studies in Environmental Policy Making
Behind the vast majority of policies lies a long and detailed decision-making process that, more often than not, includes some form of Cost-Benefit Analysis (CBA). CBA aims to estimate the market and non-market costs and benefits of a given policy proposal to better inform policymakers’ decisions to implement a policy. Of the three categories used to quantify non-market values, revealed preference, stated preference and benefits transfer valuations, this paper will focus on the stated preference method because the subject of this study cannot be easily measured by observable behaviour.
Underlying the popularity of CV studies and other CBAs in policy making is the assumption that some quantifiable value can be attached to the environment. While there are some who hold the view that environmental assets should be protected at all costs (e.g. Hargrove, 1992), this is unlikely to be a tenable position for policymakers who bear a much broader responsibility for the well-being of their states and peoples. Instead, policymakers are likely to take a more nuanced approach in determining their policy priorities, and accordingly, the number of resources allocated to each cause. This requires a uniform means of assessing policies to facilitate the evaluation and comparison of their policy options. Consequently, non-market valuation techniques were developed to value all goods in terms of money because it is finely divisible and an exceptionally effective substitute for most goods (Bateman et al., 2002). The yardstick of choice for most policymakers is the test of economic efficiency, meaning that resources will only be allocated to a policy if the gains from the reallocation exceed the losses. CV studies are thus crucial to environmental policymaking because in addition to allowing policymakers to value a non-market good, the environment, it provides justification for the continued existence of the environment using quantifiable economic arguments concerning society’s welfare.
CV studies can be tackled from two angles, Willingness-to-Pay (WTP), where respondents are asked how much they are prepared to forego to secure a benefit or avoid a cost, and Willingness-to-Accept (WTA), which evaluates how much compensation one must receive to forego a benefit or to suffer a cost (Hanemann, 1999). Although in theory, the two measures should yield the same result, there is an extensive body of research that suggests the two diverge in value, with WTP being between up to 2 to 5 times greater than WTA values for the same good (Pearce and Markandya, 1989). Some possible reasons for the divergence include: errors in the research methodology (Garrod and Willis, 1999); ‘endowment’ effects (Kahneman et al., 1991) and loss aversion (Kahneman and Tversky, 1979) [see: Kolstad and Guzman (1999) for a detailed discussion]. Fascinating as this body of literature may be, this study will not be going into this discussion. Instead, it will utilise WTP because WTP questions are more intuitively understood by respondents (Bateman and Turner, 1993) and the divergence between the two should be minimal when concerning only a small fraction of income (Sugden, 1999; Just et al., 1982; Freeman, 1993).
1.2 Background of Global Climate Change Mitigation Efforts and the Green Climate Fund
In 2017, Climate Change was named as one of the top threats facing the modern world (Pew Research Centre, 2017). Even prior to this recognition, it has been a critical public policy concern over the last decade or so, evidenced by the number of national policies implemented in recognition of this concern. For example, in 2009, the British Government announced its aim to reduce carbon emissions by 80% compared to 1990 levels by 2050 (Committee on Climate Change, 2009). Despite these national efforts, there appeared to be a growing recognition amongst world leaders that the sum of these individual efforts would be insufficient to combat climate change, culminating in the signing of the Paris Climate Agreement at the 21st United Nations Climate Conference in December 2015. The agreement was lauded as a watershed moment in the fight against climate change and heralded a new era of global cooperation (Zheng et al., 2016).
Although the signing of the Paris Agreement is undoubtedly a promising sign, there remain some fundamental issues with the Agreement that may undermine its ability to achieve its goal of keeping global temperature increases below 2°C. As part of the Agreement, the Green Climate Fund (GCF), a global fund set up by the United Nations Framework Convention on Climate Change (UNFCCC), was named as a critical financial pillar. Notwithstanding the United States’ recent withdrawal, however, the voluntary nature of the Agreement is a large potential obstacle to its success. Critically, given that all contributions to the GCF are voluntary, member states have full autonomy to decide what amount they wish to pledge to contribute to the GCF in any given year, and whether they ultimately fulfil their non-binding pledges. This means that while significant, the advanced economies’ agreement to jointly mobilise US$100 billion annually by 2020 as part of the Paris Agreement should not be prematurely celebrated.
Without looking too far in the past, we can easily find several instances of “landmark” international environmental agreements ultimately resulting in failure, with the Kyoto Protocol being perhaps the most famous example. Boulet et al. (2016) argue that this decline in the development of a global environmental order is due to the onset of multilateralism – it is no longer sufficient to have a select few states leading the way. As the world is minimally no less multilateral today than it was when the Kyoto Protocol was signed, it remains to be seen whether the early optimism surrounding the Paris Agreement and the GCF will translate into meaningful results (Clémençon, 2016). With the potential for failure in mind, it is imperative that we seek out means to improve its chances of success to avoid the detrimental and irreversible consequences that will arise from another failure to mitigate Climate Change.
1.2.1 The Green Climate Fund (GCF)
The GCF was set up in 2015 to leverage on the contributions from its 194 member states to support developing countries’ efforts to mitigate and adapt to Climate Change (GCF, 2017). At present, the GCF is running 54 projects worldwide in Africa, Asia Pacific, Latin America,
the Carribean and Eastern Europe. The projec
ts are expected to increase the resilience of 159 million individuals to Climate Change and will lead to the avoidance of 1 billion tonnes of carbon dioxide. The GCF’s projects fall into three categories: mitigation, adaptation, and cross-cutting projects, of which 40% of funding goes to mitigation efforts (GCF, 2017). Generally, investments are channelled into projects seeking to promote ‘low-emission, carbon resilient’ development in states most vulnerable to the effects of Climate Change. Beyond simply relying on contributions from the public sector, the GCF also uses the contributions from their member states to stimulate private financing, in recognition of the scale of the Climate Change problem.
As of September 2017, the UK has pledged an average of £15 per citizen, of which it has paid about £8.33, or 56%. The UK’s total pledge was the third highest in the world, after the US and Japan, while its per capita pledge was the fifth highest, behind Luxembourg, Sweden, Norway, Finland (GCF, 2017).
1.2.2. Domestic Climate Change Mitigation Efforts
In October 2017, the Conservative Party announced their new Clean Growth Strategy which aims to deliver increased economic growth with decreased carbon emissions (HM Government, 2017). Some key proposals of the strategy are summarised in the table below:
While relatively ambitious, doubt remains over whether the standards being pursued under this strategy are sufficient to bring the UK in line with the Paris agreement’s goal of a maximum of 2°C increase in temperatures from pre-industrial levels (HM Government, 2017; CCC, 2016). This is concerning given the enormous costs incurred over domestic policies, as seen from the table above, relative to the pledge to the GCF. However, we must bear in mind that the aims of the two are different. Although the Clean Growth Strategy takes into account the need for environmental awareness, it ultimately prioritises economic growth. Hence, the disparity in expenditure does not by any means suggest that contributions to the GCF will be inconsequential.
1.3 Aims and Objectives
The aim of this paper will be to evaluate WTP measures for contributions to a non-market good under different levels of information provision. The information effects will be measured in two ways. Firstly, we will assess the effect of access to information on WTP by comparing respondents’ answers prior to and after receiving additional information. Next, we will assess how WTP responds to access to information regarding the payment vehicle by comparing the change in WTP across the different treatment groups.
A secondary objective of the paper is to evaluate the WTP of individuals in the UK for the GCF and to analyse the reasons underlying their WTP. This has wider implications for the success of the GCF and international environmental agreements in general, particularly in informing governments of the most effective way to align their contributions to such efforts with the preferences of their populace. Ensuring that government actions are aligned with what the people want is especially important in sustaining international funds like the GCF since public opinion often directs public policy (Burstein, 2003). Given that we will likely see an upswing in international climate agreements as the effects of Climate Change become more apparent, the results of this paper, and particularly the insights it may provide with regards to influencing domestic support levels for such agreements, are likely to be of key significance in the years to come.
The paper will adopt the following structure. Section 2 will examine existing literature about information effects, payment vehicles, tax reallocation and information levels. Next, Section 3 will explain the crucial elements of the study design while section 4 will present the results of the study. Following which, Section 5 will outline the limitations of the study and finally, the policy considerations will be discussed together with the conclusions of the study in section 6. As far as I am aware, this paper is the first to attempt a valuation of contributions to the GCF.
2. Theory and Existing Literature
2.1 The Importance of Information Effects in CV Studies
‘Information’ can be defined as “the stock of knowledge the individual has of the true state” (Munro & Hanley, 1999). Since we live in a world of imperfect information, there will always be an element of uncertainty surrounding the decisions individuals make. As established by Graham (1981), uncertainty in the world results in consumers having multiple measures of value depending on the context in which the CBA is made. Consequently, information effects are exceptionally important because they set the context within which respondents value the subject of the study, and can have profound impacts on the result. All CV studies are thus subject to an “information bias”, where the information provided will shape a respondent’s formulated WTP.
Information bias may be introduced into CV studies by affecting the probabilities attached to different possible benefits, improving the credibility of the CV process, and altering the link between the individual’s formulated and revealed WTP where responses to surveys may be strategically biased (Munro and Hanley, 1999). Boyle (1989) also suggests that new information affects measures of dispersion, marginally adjusting the WTP of those who are well-informed while substantially adjusting the valuations of the “ignorant”, which should narrow the gap between the WTP of the two groups. In line with Bayes’ Theorem , the interaction between the respondent’s prior knowledge and the new information provided will also affect the results (Hirshleifer and Riley, 1992). This interaction between prior knowledge and new information is especially pertinent for non-market goods where the market is not able to intervene and provide signals through prices as it may sometimes do (Grossman and Stiglitz, 1981). This implies that individuals will have different responses to new information and that we should not expect the provision of new information to necessitate an increase in formulated WTP for the individual or society as a whole (Munro and Hanley, 1999). Consequently, information effects are particularly significant in CV studies eliciting the non-use values of environmental amenities (Munro & Hanley, 1999).
That being said, over the years, a considerable body of empirical evidence has developed to show that while no generalisations can be drawn about new information as a whole, the provision of certain types of information can bring about significant changes in WTP. For example, in their study of mean WTP for prime-land preservation in the US, Bergstrom and Dillman (1985) found that the mean WTP was significantly higher in the group of respondents who were given information about the scenic and environmental benefits from preservation than in the uninformed group. The studies also found that in general, the stronger the use value for the good, the less sensitive these goods’ values are to new information as respondents had more personal experience in dealing with the good (Bergstrom et al., 1989; Boyle, 1989). This suggests that it is possible to draw conclusions about the effects of particular types of information on WTP. Other notable studies and their findings are summarised in the table below:
2.2 The Role of Payment Vehicles in CV Studies
Although some attempts have been made to compile best-practice “manuals” of CV studies (e.g. Bateman et al., 2002; Arrow et al., 1993), there are still many unresolved issues surrounding the most effective means of administering CV studies. This is part
icularly pertinent given that, as intimated above, the v
aluation process, and thus its resultant values, are dependent on the CV scenario (Jakobsson & Dragun, 1996) –valuation cannot be separated from the wider context within which it is placed (Mitchell & Carson, 1989). One such issue is the payment vehicle (PV) used in the study. Some more commonly-used PVs include various forms of taxation, entry fees, changes in utility bills and trust fund payments (Garrod & Willis, 1999).
The choice of PV presented in the CV study is important because it is a key factor setting the context for respondents and can thus have great impacts on the reliability of WTP values (Morrison et al., 2000). There is a large body of empirical evidence that supports the significant influence PV has on the outcome of the CV (Green & Tunstall, 1999). For example, Jakobsson and Dragun (1996) suggest that if respondents perceive the chosen PV as unrealistic, they may not present a true bid. Furthermore, Bateman et al. (1993) found, that using taxation as a PV produced more consistent responses and resulted in significantly fewer zero responses than donation mechanisms. They suggest that this could be due to the incentive to free ride introduced by donation mechanisms. Consequently, it is critical that the chosen PV is credible and realistic. It should also be aligned with existing institutions respondents would be familiar with and the cultural context of the study area (Jakobsson and Dragun, 1996). Other suggested criteria for effective PV methods include the perception of being “fair” and “equitable” relative to those benefiting from the policy (Garrod and Willis, 1999), and using only methods perceived to have a direct link to the provision of the resource in question (Hanemann, 1995).
2.3 Tax Reallocation as a Payment Vehicle
The use of tax reallocation as a PV was first formally proposed by Bergstrom et al. (2004). Prior to this, CV studies tended to utilise special taxes, as recommended by the National Oceanic and Atmospheric Administration (NOAA) Blue-ribbon CV Panel (Arrow et al., 1993). Bergstrom et al. (2004) argued that tax reallocation was a more favourable PV because, in practice, governments were less likely to impose politically unpopular taxes to finance non-market goods. Instead, they were more likely to reallocate existing tax revenue to the provision of the good if it was seen as being sufficiently important. Furthermore, they also suggest that tax reallocation is a more desirable payment vehicle for governments from an equity standpoint since they found that welfare measures corresponding to a tax reallocation may not be influenced by nominal income.
Bergstrom et al. (2004) also found that because it changed neither the individual’s disposable income nor the price of the good being valued, tax reallocation led to a greater WTP amongst respondents than a special tax, implying that respondents had a greater marginal value of disposable income than of the provision of other public goods. In addition to the findings of Bergstrom et al. (2004), there has also been some empirical evidence to support the appropriateness of tax reallocation as a PV. For example, a study on the WTP for installing an environmentally friendly drainage pipe by Morrison et al. (2000) found that 32% of the pre-test sample believed that the government should pay for the project out of existing funds.
Tax reallocation also avoids some of the pitfalls of alternative PV methods such as coercive or voluntary vehicles and commodity price changes. For example, where a commodity’s price is altered, Bateman et al. (2002) argue that respondents can adjust to the higher prices by adjusting the quantity consumed, suggesting that the derived CV value will be an inaccurate reflection of real life. Furthermore, numerous studies (e.g. Carson (1997); Hanemann (1995); Garrod and Willis (1999)) have found voluntary vehicles to be inadequate elicitation methods because they incentivise free-riding and other forms of strategic behaviour . Such mechanisms may also be seen as inequitable or ineffectual in delivering the desired outcome (Hidano and Kato, 2002). Finally, the effectiveness of coercive vehicles like taxation depend strongly on the socio-economic and political context. Where there are high instances of corruption and low levels of trust in the government, the use of taxes as the PV may well increase the instance of protest responses and affect the WTP (Green & Tunstall, 1999). Where governments are particularly responsive to CV studies, WTP may be understated by individuals fearing that their taxes will be raised based on the results of the study (Green & Tunstall, 1999).
2.4 Theoretical Framework of Tax Reallocation
The following theoretical framework for tax reallocation is based on the work of Bergstrom et al. (2004) and Kontoleon et al. (2005):
Let u = u(X,Q,Z) be a household’s preference function, where u is a direct utility function, X represents private market goods, Q represents the public good being investigated, in this case the GCF, and Z is the composite commodity representing all public goods except Q, with the unit price and value equal to the tax levied on each household. Next, we assume that all households are utility-maximising, subject to their budget constraint M = PX + Z, where M represents nominal income and P is the price vector of private goods. Since Z is equivalent to the taxes paid, we assume that each household spends all of its disposable income, M – Z, on PX.
From these assumptions, we can derive the total expenditure on private and public goods required to achieve u given P, Q and Z as related to the conditional expenditure function (see: Bergstrom et al., 2004).
e = e(P,Q,Z,u) = e*(P,Q,Z,u) + Z ———- (1)
where e is nominal income and e* is disposable income.
The change in GCF contributions is then introduced through a change in Q. Assume that the benefits from the present level of GCF contributions can be represented by Q0 and will improve to Q1 over time, where Q1 > Q0. If there is no change in nominal income and we assume there to be no direct or indirect trade-off in household expenditure, this change in Q would lead to a change in Z. This trade-off is named by Bergstrom et al. (2004) as Compensating Tax Reallocation (CTR) which is calculated by:
CTR = Z0 – Z1 ———- (2)
In other words, CTR is equal to the marginal rate of substitution between the GCF and the composite commodity of all other public goods.
2.5 Level of Information
The focus of this study will be on evaluating the impact of additional information concerning tax reallocation on the WTP. As asserted by Boyle (1989), variations in information will produce different results. For example, asking someone to value a generic house will yield a very different result from asking the same individual to value a house given its condition, size and location. This raises two key questions, what is the optimal level of information that should be provided, and what type of information should this be?
As with any economic good, optimality is defined as the point where marginal benefits are equivalent to marginal costs. Beyond that point, the marginal costs will exceed the marginal benefits, resulting in a net welfare loss. In the context of information, the marginal benefit of information is zero where extra information will not change the decision that has been made. On the other hand, the marginal costs of information refer to the time, effort or resources required to gain the extra information. However, where public goods are concerned, there tends to be a divergence between the
social and the private value placed on information, and thus we tend
to have a sub-optimum actual level of information. This divergence can arise in several ways in CV studies.
Firstly, respondents in CV studies may not be primed to make decisions about the subject of the study, especially for environmental goods that have little use value to the individual. Without any prior personal experience in trading the good, the individual may attach very low personal relevance to it and by extension, any information regarding it. The private value of such information would thus tend towards zero. For example, information about a national park halfway around the world may be of little value to an individual given that the potential utility the individual can derive from it is minimal. This is described as the distance decay effect (Bateman, 2006). On the other hand, the social value of information concerning the same park would be considerably greater given the positive externalities it brings about simply by existing, such as aesthetic or environmental value.
Next, divergence arises where the good in question is a public good valued through group decision by individuals with heterogeneous preferences (Munro and Hanley, 1999). Within a heterogeneous group of individuals, there will always be a wide range of possible valuations. As the size of the group increases, the impact of an individual’s valuation will diminish. In recognising this, the private value placed on the information will tend towards zero since the likelihood that obtaining any new information will change the social outcome decreases with the size of the group.
Finally, the non-rivalrous nature of information means that once one party has spared the expense of acquiring the information, the marginal cost of other parties acquiring the information is zero. They may acquire the information either directly from the source, or indirectly by observing the behaviour of informed parties. This creates a strong incentive to free-ride which, under free market conditions, would result in no expense being spent on information acquisition.
These problems suggest that respondents will almost universally value information below the socially optimal level. Hence, while it may not be possible to establish a socially optimal level of information given the heterogeneity of preferences, a higher level of information will almost certainly bring respondents closer to the socially optimal level of information (Jakobsson and Dragun, 1996). That being said, caution must be exercised to avoid providing excess information. Sagoff (1988) suggests that, given the upward-sloping nature of the marginal cost curve for acquiring further information, the optimal social level of information will be below perfect information, where marginal social value cuts the marginal cost curve at Q*, as shown in Figure 1. Consequently, the study should strive to provide more information to the extent that it can be acquired at a reasonable cost and included in the study in an easily comprehensible manner (Jakobsson and Dragun, 1996).
3. Survey Design
This study can be broken down into three components: demographic, attitudinal and knowledge questions; pre-treatment assessment of WTP for the GCF; and the post-treatment WTP survey.
The first component will ask for demographic details of respondents, namely sex, age, ethnicity, income level, level of education and which region of the country the respondent lives in. Attitudinal questions will assess respondent’s attitudes, such as what they believe are the most key policy areas, and whether they believe Climate Change is a legitimate threat to humanity . Finally, knowledge questions will assess the extent of their existing pool of knowledge regarding the Paris Agreement and the GCF. This list of variables is adapted from that used in Berrens et al. (2004)’s study of information effects on WTP for the Kyoto Protocol, and Hanley et al. (1995)’s study of Biodiversity valuation.
The remaining two components will adopt the WTP survey format outlined in Green & Tunstall (1999). Their work states that a WTP survey comprises of five components: the introduction, the PV, a filter question on whether the respondent is prepared to pay anything at all, an elicitation method to determine how much those who are prepared to pay are prepared to offer, and a follow-up question asking respondents for the reasons why they made such a decision. We will now examine the elements of the survey in detail.
3.1 Background Introduction to Respondents
In line with the literature discussed above, the introduction must inform respondents that a refusal to pay due to a lack of resources or preference for the GCF is a legitimate response to the survey. This complements the filter question and minimises the chances that respondents may feel pressured to give a positive WTP because they believe that to be the ‘preferred’ result (Arrow et al., 1993).
Next, respondents must also be reminded to take the study as if it has real impacts on their own resources, taking current and actual resource constraints into consideration. This will serve to minimise the occurrence of hypothetical bias, which arises from the understanding that the referendum is hypothetical, leading to inaccurate representations of WTP. Over the years, numerous studies have found the existence of hypothetical bias in both private and public goods in studies conducted with all manner of bidding mechanisms (Brown et al., 2003), suggesting that it is a problem to be expected in any CV study. Consequently, it is imperative that steps be taken to minimise the interference of hypothetical bias with the results of the survey. Brown et al. (2003) found that by including reminders of the need for careful consideration and informing respondents of their susceptibility to hypothetical bias, the incidence of such bias is minimised. These reminders have been found to help respondents to focus on answering the question they are given rather than using their response to send a message. The results are in line with earlier findings by Loomis et al. (1996), Cummings and Taylor (1999) and List (2001).
The introduction of new information will always be followed by several follow-up questions to ensure that respondents have read and understood the information provided. The information provided to all respondents include information about the GCF, Climate Change and how much the average British citizen is already paying towards the GCF. While somewhat controversial because of the anchoring effects the latter will introduce (Green et al., 1993b), such anchoring effects may be necessary since respondents are not familiar with valuing environmental protection and may thus need a clearer gauge of what concrete results arise from their contributions. Further, Schläpfer & Schmitt (2007) argue that all choices, whether about private or public goods, tend to be subject to some form of anchoring effects. Thus, instead of attempting to eliminate such effects, we should instead provide informative anchors, such as the current average amount being paid, to respondents. Depending on the treatment group, the survey will also introduce the PV.
Finally, a filter question will be included to distinguish respondents with a positive WTP from those without. Respondents who give a zero-answer will be asked to explain why they made such a decision, including whether their response was given as a ‘protest vote’, meaning that the respondent does not believe monetary values should be attached to the good in question (Green and Tunstall, 1999). On the other hand, those with a positive WTP will be routed on to the elicitation question to determine their WTP. With each additional piece of information introduced, respondents will be asked the filter q
uestion again and complete the valuation exercise
if their WTP is positive.
3.2 Referendum Elicitation Format
Since environmental CV studies are valuing subjective goods or services, purely descriptive responses would not provide for much meaningful analysis, especially given the abundance of evidence to show how changes in the context and wording will affect the results (Green & Tunstall, 1999). In order to maximise the reliability of CV estimates, the NOAA panel recommends that CV studies be conducted in a referendum format (Arrow et al., 1993) because it best approximates the real-life referenda that are taken to be representative of “true” preferences in policymaking. Furthermore, there is no incentive the engagement of strategic behaviour in a referendum. The referendum format entails providing information to the respondents and then asking them to answer simple “yes/no” questions. For example, “Would you be willing to contribute $X from the amount you currently pay in taxes to the Green Climate Fund?”
This study will utilise a triple-bound dichotomous-choice referendum format, as shown in figure 2. Although there are other alternatives, such as open-ended CV questions , the panel concludes that open-ended questions are unreliable because the scenario presented to respondents is unrealistic. Furthermore, open-ended questions may invite the strategic overstatement of a WTP or WTA compensation to “make a point” (Arrow et al., 1993). While there are some problems with the referendum format such as inaccurate answers due to the perceived feasibility of the proposed action, the inclusion of questions to test for such biases, which are well-documented in the literature, can minimise their impact on the results (Arrow et al., 1993).
.3 Treatment Methods and Hypothesis
Using a split-size referendum, the study will comprise of four treatment groups:
The differences in treatment are adapted from a study by Nunes and Travisi (2009) that compares the impacts of tax reallocation from expenditure on public transport and that of expenditure on administration on WTP. Although that study concluded that the marginal value of public money did not appear to depend on the budget source as the WTP between the two sources were not statistically significant, this result is ultimately constrained by the fact the respondents did not have a choice in the budget source. Hence, the purpose of treatment group IV is to assess if choice of source has any influence over WTP.
Accordingly, the hypotheses of this study are formalised in the following three statements:
Hypothesis 1: The valuation of the GCF is dependent on the introduction of further information to the respondents
Hypothesis 2: The valuation of the GCF is dependent on whether the payment vehicle is a special tax or a tax reallocation scheme.
Hypothesis 3: The valuation of the GCF is not dependent on whether respondents have a say in the source of the funding.
Following the methodology adopted by Bergstrom et al. (2004) and Nunes and Travisi (2009), we will conduct a two-tailed analysis on the results to test each hypothesis.
5. Limitations of the Study
Sagoff (1988) argues that by providing information beyond what is originally known to the respondent, their valuation becomes endogenous to the valuation process as this information would not have been available to them in a real-world situation. This casts doubt on whether the results are truly representative of respondents’ WTP. That being said, since a secondary aim of this study is to understand how policymakers should best craft information campaigns to bring public opinion in line with their policies, the fact that valuation may be endogenous should not be seen as a flaw of the study. Rather, it amplifies the importance of understanding such information effects so that policymakers may use this endogeneity, if it truly exists, to their advantage.
Another limitation of the study arises from its medium as an online survey conducted by a survey company, as was the case in Berrens et al. (2004). This meant that respondents were limited only to those comfortable with Internet usage. However, given the high level of connectivity in the UK, the limitation of internet access is minimal in this context. There was also a potential for selection bias since the respondents were chosen from a pre-recruited respondent panel. Ultimately, it is difficult to eliminate some kind of selection bias from any pool of respondents, and this methodology was chosen because it allowed for respondents to be more representative samples of the UK’s demographic, provided for anonymity, eliminated interviewer bias, and was relatively cost and time efficient (Bateman et al., 2002).
Since the study is conducted as a one-off internet survey, it will not be able to account for the “time-to-think” effect. This was found to be significant in Hanley et al. (1995) and Whittington et al. (1992) where respondents were found to revise their initial bid after having some time [two weeks, in Hanley et al. (1995)] to reflect on their answers. However, this re-surveying was not viable given the resource and time constraints faced.