not an income-neutral preference scale, because the premise is one dollar, one vote. Different groups in society have varying amounts of dollars or income that affect their willingness to pay.
This discussion shows that benefit estimation is directly connected to the economic concept of demand. The goal of benefit measurement is to estimate the demand and, thus, the willingness to pay for the output of the project. Therefore, all the empirical demand-estimation issues discussed in chapter 5 are relevant here. These include the problems of incorporating all relevant vari ables influencing demand and holding their effects constant while examining the relationship between price and quantity demanded.
Willingness-to-pay issues become more complicated in cost-benefit analysis, however, because many of the project outcomes being evaluated, such as good health or a clean environment, are not bought and sold in mar kets. Furthermore, prices may not be charged for many government-provided goods and services either because it is impossible to do so (the public goods problem) or because society has made a conscious decision not to do so (the income distribution or equity problem). Therefore, willingness to pay inthese cases must be estimated, either through observation of behavior in hypotheti cal or contingent markets, or through inference from indirect methods on behavior in markets related to the project outcome in question. Thus, the choice is between methodologies that focus on how individuals answer ques tions about how they would behave in certain situations, and methodologies that infer willingness to pay from observing the behavior of individuals in markets affected by the relevant project outcome. These methods will be summarized and catalogued under several broad headings below. More detailed descriptions of all of thest! methods, particularly in relation to the health and environmental areas of interest in this chapter, can be found in Cummings, Brookshire, and Schulze (1986); Gramlich (1981; 1990); Free man (1993); Tolley, Kenkel, and Fabian (1994); Portney (1994); Hanemann (1994); Diamond and Hausman (1994); Bjornstad and Kahn (1996); O'Brien and Gafni (1996); Smith (2003); Haddix, Teutsch, and Corso (2003); Bayoumi (2004); and Drummond et al. (2005).
Estimation of willingness to pay from surveys and other hypothetical situ ations is known as the contingent valuation method. This is the method most closely related to the direct estimation of a market demand curve, except that the market is hypothetical. Contingent valuation methods ask individuals to reveal their personal valuations of increases or decreases in unpriced goods, • either through surveys or in experimental situations. Individuals are given information about the good in question, the institutional structure under which it will be provided, the method of payment, and the decision rule for