Choice Between WTP and WTA

We can easily see that in a market situation WTP and WTA are logically equivalent to the buyer's best offer and the seller's reservation price, which are Pareto-safe prices. Therefore, the use of the potential Pareto improvement (PPI) criterion actually requires that beneficial and adverse effects have to be valued in terms of compensating and equivalent measures of welfare change (Randall 1987, p. 245). The policy implication is that if a compensation test allowing intemersonal cardinal utility comparisons is regarded as a proper criterion for social decision-making, then WTP and WTA expressed in equivalent and/or compensating measures (variation or surplus, depending on the nature of the change) are theoretically the most correct measures of welfare change.

Apparently, the choice between WTP and WTA measures is in a decisive role when an elicitation question in a CVM survey is phrased. Unfortunately, their use in the case of environmental and other nonmarket goods is not as straightforward as it is when market goods are in question. The nonrival and nonexclusive nature of environmental goods makes it impossible to define property rights in a nonattenuated manner. This complicates the analysis to some extent and, as a consequence, one of the most enduring controversies in the field of CVM research is whether the WTP or WTA approach should be chosen.

Willig (1976) showed that, in the case of a price change, the difference between WTP and WTA was a function of income elasticity. Moreover, he showed that in the case of the income elasticity having a reasonable value, the difference between WTP and WTA had to be small. However, in many empiri-cal studies CVM researchers continuously reported considerable differences between WTP and WTA measures for the same good being valued. It appeared that his theoretical results were mainly valid for policies that changed prices for consumers with well-behaved utility functions. Above ali, this implied that his results might not be appropriate for some commonly estimated but not conven-tional demand functions. This is true especially in cases where the provision of public goods involves quantity changes and, in particular, for policy changes where the quantity of the good provided goes to zero (Mitchell and Carson 1989; p. 31).

Randall and Stoll (1980) expanded Willig's (1976) analysis from price changes to a situation where a quantity or quality change of a commodity alters welfare.

They observed that the CV and EV are identical and equal to Marshallian consumer surplus if commodities are fully divisible and tradable without trans-action costs in an infinite market. For fully indivisible commodities CV is larger than EVin a welfare loss case and smaller in a welfare gain case. In addition, the difference between CV and EV is somewhere in the middle of these two cases if the commodity is divisible and if its exchange claims transaction costs. As a

practical guideline they concluded that if the commodity is divisible, the com-modity market is fairly competitive, the transaction costs are low, and the price elasticity is small, then the use of Marshallian consumer surplus causes only a small error when compared to possible inaccuracies in estimation techniques. At the same time, CV and EV are so close to each other that WTP and WTA measures should be almost equal.

Hanemann (1991) showed that in the case of imposed quantity changes, the theoretical difference between WTP and WTA is governed by the ratio of the income elasticity to the• substitution elasticity rather than by the income elastic-ity alone. The substitution elasticelastic-ity refers to the ease with which other market commodities can be substituted for the given public good while maintaining an individual at the constant level of utility. The elasticity of substitution takes the value of zero if no amount of increment in any market goods can substitute for the change in the public good, and the value of infinity if at least one market good is a perfect substitute for the public good. It is possible to show that the smaller the substitution effect and the larger the income effect is, the greater is the disparity between WTP and WTA. Furthermore, if either the income effect is zero or the substitution effect is infinite, WTP and WTA must coincide. If the public good in question is unique and the income elasticity is of ordinary magnitude, the difference between WTP and WTA can become considerable.

There is no doubt that Hanemann's (1991) work is unsettling because it implies, in contrast to Willig' s (1976) results, that there may exist large real differences between WTP and WTA measures for unique environmental goods. This sug-gests that the property rights chosen are of great importance.

Shogren et al. (1994) tested Hanemann's (1991) propositions. The authors analyzed experimentally in a non-hypothetical auction market situation differ-ences between WTP and WTA by using easily substitutable market commodi-ties (i.e. candy bars and coffee mugs) as well as a nonmarket commodity of low substitutability (i.e. reduced health risk). They found out that WTP and WTA amounts related to market commodities with close substitutes did not differ statistically significantly from each other. The opposite was true in the case of a nonmarket commodity without close substitutes. The WTP and WTA measures diverged and the difference persisted even with repeated market participation and full information on the nature of the good. Thus, the results of Shogren et al.

(1994) give support to Hanemann's (1991) hypotheses.

The most serious problem in CVM studies using WTA type questions is that it is difficult to create plausible contingent market scenarios (Carson 1991).

Consequently, the current consensus among CVM researchers seems to be that WTA cannot reliably be measured by using a CVM survey. Already Cummings et al. (1986, pp. 102-106) in their "reference operating conditions" for contin-gent valuation studies recommended that only WTP measures should be elicited in the attempt of valuing environmental commodities. In addition, the recom-

mendation of the NOAA Panel (Arrow et al. 1993) is that WTP format should be used instead of WTA format because WTP is a more conservative choice.

Generally, when aspects of the survey design and the analysis of responses are ambiguous, the option that tends to underestimate the acquired welfare measure is preferred. However, the NOAA Panel did not review any theoretical issues concerning the superiority of WTP to WTA.

As noted above, the formulation of the elicitation question also depends on the prevailing set of property rights. It does not make sense to present a WTA question to a non-owner of a resource if the person cannot simultaneously be guaranteed property rights that ensure his entitlement to the compensation of-fered. Mitchell and Carson (1989, pp. 38-41) have developed an interesting approach that rethinks the role of property rights when selecting between WTP and WTA in the case of public goods which require annual payments or their equivalents in order to maintain the given level of quantity and/or quality. Many important public goods have this feature. Water quality, for instance, would rapidly decline if the government did not spend any money on control measures.

For public goods of this type, neither ownership nor use as such is able to capture the relevant relationship between the good and the consumer.

Two dimensions of a public good receive special attention: it is held either individually or collectively, and its level of quantity/quality is either currently accessible or currently inaccessible. Individually held rights to a public good can occur when a public good is clearly excludable. Then individuals can be granted exclusive rights to its use by the collective if such a grant is considered to serve the public interest. Various allocation mies are used when the rights are granted, and the rights are usually not transferable. Collectively held rights occur when access to the public good is available to ali members of the collec-tive, and when individual members cannot trade their right to access. The rural environment is a good example of a public good of this kind. People have a collective property right to visit the countryside (public access), but the right cannot be transferred by any means. Now, there is a cost to provide the good at a given quality level, and this cost is borne by ali consumers through some combination of taxes (agricultural support), prices (agricultural products), and the like. Moreover, if the level of payment is not maintained, the quality of the rural environment will soon deteriorate. Correspondingly, if a quality increase of the rural environment is desired, higher payments (higher taxes, higher pro-ducer prices) will be needed to cover the cost of providing the new quality level.

The second dimension for determining the appropriate surplus measure for a public good is whether a given level of quantity/quality is accessible. At the moment, there exists a certain number of exceptionally valuable traditional agricultural landscape areas that have been tracked down by governmental authorities. This can be regarded to represent the quality level of the rural environment. On the other hand, it is currently impossible to have twice as many

exceptionally valuable traditional agricultural landscape areas. This framework for conceptualizing the property right to public goods has important implica-tions for the choice of the correct Hicksian surplus measure for CVM surveys.

Now, the aim is to measure the benefits of this public good starting from the consumer's initial level of utility. If a given utility level of the public good is not currently available, the WTI'cs measure determines the would-be value of the increased provision. This is parallel to a private good case in which an indi-vidual has similar property rights (i.e. he is not able to use the good and does not own it). In both cases, the WTPcy correctly measures the amount that the individual is willing to pay for the improvement that leaves him as well off before the change as after it.

In the public good case, however, the WT13cs is also regarded to be the most correct measure for a proposed decrease when a given quality level is currently available. The corresponding set of property rights for a private good case concerns a situation where an individual has access to a private good (i.e. he owns the good and is able to use it). We can easily observe (see Table 4.1) that in this private good case (quantity decrease) the recommended welfare change measure is WTAG5 instead of the proposed WTI)cs if the measurement is to be carried out starting from the initial level of utility. Now, since the individual in the public good case is already paying for the public good on a regular basis, the Hicksian compensating surplus for this case is the amount that the consumer is willing to pay to forgo the reduction in the quality of the good and still be as well off as before. If the rural environment is thought of as an example in this context, people have two options: better quality of the rural environment and higher payments or poorer quality of the rural environment and lower payments.

In the case of quality decrease of the rural environment, the WTA question would inquire how much people would have to be paid to voluntarily accept the poorer quality of the countryside. However, as explained above, people are already financing the maintenance of the rural environment through taxes. In such a situation, people have the right to enjoy the quality of the rural environ-ment but, simultaneously, they have to pay for it. Thus, the effective property right is actually WTP, not WTA. The essential property which makes WTP rather than WTA appropriate is that the same people constitute both sides of the transaction.