Seminar - Discussion

	Contingent Valuation v Stated Preference
	The Definition of Consumer Surplus
	Future Research
	Multiple Responses
	Household Representative

Contingent Valuation v Stated Preference (Conjoint Valuation)

Both techniques have a history of actual surveys and associated research. Each method is appropriate in some circumstances. Stated preference survey methods obtain best results when the attributes of a good can be clearly defined and the objective of the survey requires masking to avoid policy bias. Some delegates advocated contingent valuation, others pointed out that contingent valuation had been shown to produce unsatisfactory results in transport studies. The rejoinder was that the form of the contingent valuation questions was important and if these had been different the current studies would have shown that the contingent valuation estimates produced comparable results. It was then pointed out that these more sophisticated contingent valuation questions, dichotomous choice, would have resembled stated preference questions. The corroboration and rescaling of SP (Hypothetical choice) studies with RP (Current choice) results is advised if the accuracy of the estimated demand function is pertinent.

The Definition of Consumer Surplus

For the individual, consumer surplus is defined as the area under the Marshallian demand curve above the current price. The Marshallian demand curve and the Hicksian demand curve coincide when the income effect is zero. This means that the areas of compensating variation, equivalent variation and consumer surplus also coincide. This is a useful result because consumer surplus can be calculated as compensating variation. Calculating compensating variation is convenient because a demand curve need not be estimated. Compensating variation can be calculated from the indirect utility function.

Future Research

The surveys had shown very little evidence that income differences affected the measures of willingness to pay. Without such income effects the different measures of welfare would be equal. The seminar participants therefore reached a consensus that it would be more efficient to devote resources to resolving issues of aggregation than trying to measure differences between equivalent and compensating variation and consumer surplus.

Multiple Responses

In the Stated Preference method a number of questions are put to the same respondent. This can give rise to bias because of potential correlation between the sets of choices made by an individual respondent; this can alter both the coefficients and the standard error estimates of the model. Correction to the standard error estimates is the most notable effect, and can affect the significance of variables in the model. If time and resource constraints permit it is advisable to perform such correction. At the seminar no consensus was obtained on what methods were suitable. However literature is available on the use of Jackknife (Cirillo, Daly, Lindveld) to solve this problem, and the methods of White and Train are often quoted.

Household Representative

The survey of TV viewers had interviewed the people who paid the TV licence. In future the questionnaires needed to determine whether the respondent was answering on their own behalf or on behalf of the whole household. The view taken in the survey was that the respondent answered on their own behalf. Their answer was probably representative of other adults, but not of children since children would not have the means to pay for television services.