The application of laboratory experimental economics to the contingent valuation of public goods

We have implicitly argued in the last section that a dynamic iterative survey mechanism may well need to be employed in the design of contingent valuation survey instruments in order to improve the accuracy of responses. Furthermore, due to the current inaccuracy of hedonic and travel cost approaches for valuing public goods, the least cost method, in our view, for establishing anchor or true individual values for testing alternative survey instruments is to use laboratory experiments. The objective of these experiments should be the development of the most simple survey design which gives accurate responses subject to the budget of the investigator. Is a complex iterative voting procedure required? How fast will such a procedure converge to ‘true’ values? What is the effect on incentives of relaxing the unanimity voting feature for large groups? All of these operational questions can at least qualitatively be answered in an experimental laboratory setting. This approach would allow rapid resolution of a number of problems which have developed in the application of the contingent valuation approach. First, the large difference between economic measures of willingness to accept and willingness to pay may be greatly reduced by application of demand revealing mechanisms. Any remaining difference between the two measures might then be properly attributed to psychological, ethical or other complicating factors. Second, the consistently large differences between the iterative bidding and payment card measures of willingness to pay suggests that one of the procedures might be more accurate than the other. Laboratory experimentation should be able to quickly identify the superior procedure. Third, contingent valuation studies which involve uncertainty have not proven successful. In a study of the willingness to pay to contain toxic wastes undertaken by Cummings and reported on in Schulze et al. (1983) nearly half of the respondents were willing to contribute the same amount of money for 50 percent odds of containment as for 100 percent odds of containment. Does this result indicate a failure of the expected utility hypothesis or a failure to perceive or comprehend probabilities by a large sub-sample of individuals? Or, is the survey at fault? Again the least cost approach for resolving these questions is likely to be a laboratory setting. Finally, individuals may have severe perception problems with the timing and method of payment used to collect bids for public goods. Schulze et al. (1983) report on a large divergence in the value of preserving visibility for visitors at the Grand Canyon using monthly payments in the form of higher electric utility bills to collect payment as compared to collecting higher daily entrance fees. Note that the first method hypothetically collects a regular payment on a monthly basis while the second hypothetically collects payments only if respondents visit the Grand Canyon. The first method implied an overall larger total benefit of preserving visibility than the second. Again, laboratory experiments could readily determine the relative accuracy of alternative temporal payment mechanisms. Copyright Martinus Nijhoff Publishers 1986