I'm typing up my notes for the contingent valuation method (CVM) lecture in the benefit-cost analysis course and it has made me ready to deal with something that's been bugging me for awhile*. I'm thinking it is time to drop the "valuation" part of contingent valuation and replace it with "behavior" (i.e., contingent behavior), or drop "contingent valuation" completely and go with "stated preferences."
The contingent valuation method was developed over the past 30+ years. It involves asking hypothetical willingness-to-pay questions to elicit economic values for environmental quality and other things. I used it in my 1990 dissertation and along with a bunch of other environmental economists spent most of the 1990s writing "CVM" scenarios, analyzing the data and fending off "revealed behavior" critics.
Historically CVM has involved questions such as:
- Open-ended: How much would you be willing to pay for X?
- Closed-ended: Would you be willing to pay $A for X?
Here is how I put it about 10 years ago (and published as Chapter 3 in the Handbook on Contingent Valuation):
Best practice has evolved to the point where closed-ended questions are strongly preferred and these are usually framed as a referendum in order to avoid problems such as a lack of familiarity with the type of situation (how often are you placed in a situation where a car dealer asks you how much you'd be willing to pay?). Voting in referenda is behavior and can be analyzed alongside actual votes. Another historically CVM question asks those pursuing outdoor recreation if they would still take the current trip if the trip cost was $A higher. This is a contingent behavior question and not CVM. As such it can be analyzed alongside revealed preference recreation data. We could say the same thing for housing market, labor market and averting behavior decisions.Willingness to pay v. behavioral intentions
A helpful suggestion might be to think of a contingent valuation question as a behavioral intention question, not as a willingness to pay question. A behavioral intention question asks people about behavior under hypothetical conditions: ‘Would you donate $15 if that is how much it cost to protect sea turtle nesting habitat?’ or ‘How would you vote on the sewer bond if paying off the bonds cost you $125 per year?’ or ‘How many shellfish meals would you eat if they were safe and cost $12?’ Many people can imagine how much money they would donate, how they would vote, and how much they would eat under different circumstances. Behavioral intentions are also easier to compare to actual behavior.
Willingness to pay questions ask people to speculate on how much they would be willing to pay for something, not what they would do in a certain situation: ‘Would you be willing to pay $55 for an increase in water quality to the fishable level?’ Most people aren’t used to being asked about the size of their consumer surplus triangle. Also, hypothetical willingness to pay is not easy to compare with actual behavior. Even so, sometimes the willingness to pay question is unavoidable because a realistic scenario can’t be constructed around a behavioral intention question. If this is the case, try to remember the difficulties that respondents have with hypothetical situations.
As far as I can tell there is not much more contingent "valuation" stuff being done anymore. Is it time to stop talking about contingent "valuation" and go with contingent "behavior"?
*Shameless plug: I've also been thinking about this issue as I'm corresponding with Tim and Ju-Chin Huang trying to meet the 2010 deadline for our revealed and stated preference data book.
