Title: Valuing the environment through contingent valuation.
Subject(s): NATURAL resources -- Valuation; VALUATION
Source: Journal of Economic Perspectives, Fall94, Vol. 8
Issue 4, p19, 25p
Author(s): Hanemann, W. Michael
Abstract: Focuses on the use of contingent valuation to
measure people's values for environmental resources.
Conducting reliable surveys; Objections to surveys;
Contingent valuation and economic theory.
AN: 9501060641
ISSN: 0895-3309
Full Text Word Count: 13108
Database: Academic Search Elite
Print: Click here to mark for print.
View Item: [Full Page Image] Full Page Image
-------------------
[Go To Citation]
VALUING THE ENVIRONMENT THROUGH CONTINGENT VALUATION
The ability to place a monetary value on the consequences of pollution
discharges is a cornerstone of the economic approach to the environment. If
this cannot be done, it undercuts the use of economic principles, whether to
determine the optimal level of pollution or to implement tiffs via Pigouvian
taxes or Coase-style liability rules. Sometimes, the valuation involves a
straightforward application of methods for valuing market commodities, as
when sparks from a passing train set fire to a wheat field. Often, however,
the valuation is more difficult. Outcomes such as reducing the risk of human
illness or death, maintaining populations of native fish in an estuary, or
protecting visibility at national parks are not themselves goods that are
bought and sold in a market. Yet, placing a monetary value on them can be
essential for sound policy.
The lack of a market to generate prices for such outcomes is no accident.
Markets are often missing in such cases because of the nonexcludable or
nonrival nature of the damages: for those affected by it, pollution may be a
public good (or bad). The public good nature of the damages from pollution
has several consequences. It explains, for example, why the damages are
sometimes large--only a few people may want to own a sea otter pelt, say,
but many may want this animal protected in the wild. It also explains why
market prices are inappropriate measures of value. In the presence of
externalities, market transactions do not fully capture preferences.
Collective choice is the more relevant paradigm.
This is precisely what Ciriacy-Wantrup (1947) had in mind when he first
proposed the contingent valuation method. Individuals should be interviewed
and "asked how much money they are willing to pay for successive additional
quantities of a collective extra-market good." If the individual values are
aggregated, "the result corresponds to a market-demand schedule" (p. 1189).
Thus, surveys offered a way to trace the demand curve for a public good that
could not otherwise be gleaned from market data. Schelling (1968) made a
similar point in his paper on valuing health. While the price system is one
way to find out what things are worth to people, he wrote, another way is to
ask people, whether through surveys or votes. Answering surveys may be
hypothetical, but no more than buying unfamiliar or infrequent commodities.
"In any case, relying exclusively on market valuations and denying the value
of direct enquiry in the determination of government programs would depend
on there being for every potential government service, a close substitute
available in the market at a comparable price. It would be hard to deduce
from first principles that this is bound to be the case" (pp. 143-4).
Schelling's point was not that indirect methods using market transactions
have no role, but rather that they cannot always be counted on to provide a
complete measure of value. Analysts can often capture some effects of a
change in air quality or a change in risk to human health through a hedonic
analysis that looks for evidence to property values or wage rates (Rosen,
1974). But people may also value those items in ways not reflected in wages
or property values. Similarly with averting expenditures and household
production models (Freeman, 1993), which rely on the demand for market
commodities that are complements to, or surrogates for, the nonmarket good.
If people value that good at least partly for reasons unrelated to their
consumption of the complementary private goods, those methods capture just
part of people's value--what is called the "use value" component, following
Krutilla (1967).[1] They fail to measure the "non-use value" or "existence
value" value component, which contingent valuation can capture.
An alternative is to turn to the political system, for example using
collective choice models to estimate demands for local public goods (Oates,
1994). However, Cropper (1994) suggests this is unlikely to be useful for
the environment because, in the United States, there are few cases where
local governments actually set environmental quality. Moreover, as Chase
(1968) noted, the method contains an element of circularity: a major reason
for the spread of benefit-cost analysis is legislators' desire to obtain
information on the public's value for government programs. While it may
sometimes be desirable to leave the assessment of value to the legislative
process, it is not obvious that this is always so. Measuring liability for
damages from pollution is an example. In some cases one wants to ascertain
how the public values something, and contingent valuation may be the only
way to measure this short of a plebiscite.
Ciriacy-Wantrup (1947) recognized that surveys are not foolproof. The degree
of success depends on the skill with which the survey is designed and
implemented. But it was time, he felt, that economics took advantage of
developments in social psychology and the newly emerging academic field of
survey research: "Welfare economics could be put on a more realistic
foundation if a closer cooperation between economics and certain young
branches of applied psychology could be established" (p. 1190). This finally
occurred in the 1980s, and contingent valuation came of age. Two landmarks
were an EPA conference in 1984 that brought together leading practitioners,
other economists, and psychologists to assess the state-of-the-art (Cummings
et al., 1986), and the publication of what has become the standard reference
on contingent valuation, Mitchell and Carson (1989), which puts it in a
broader context involving elements from economics, psychology, sociology,
political science, and market research.
Contingent valuation is now used around the world (Navrud, 1992; Bateman and
Willis, forthcoming), both by governments agencies and the World Bank for
assessing a variety of investments. A recent bibliography lists 1600 studies
and papers from over 40 countries on many topics, including transportation,
sanitation, health, the arts and education, as well as the environment
(Carson et al., 1994c). Some notable examples are Randall, Ives and Eastman
(1974) on air quality in the Four Corners area, the first major non-use
value study; Brookshire et al. (1982) on air pollution in Southern
California; Carson and Mitchell (1993) on national water quality benefits
from the Clean Water Act; Smith and Desvousges (1986) on cleaning up the
Monongahela River, Jones-Lee, Hammerton and Phillips (1985) on highway
safety; Boyle, Welsh and Bishop (1993) on rafting in the Grand Canyon;
Briscoe et al. (1990) on drinking water supply in Brazil; and the study on
the Exxon Valdez oil spill I helped conduct for the State of Alaska (Carson
et al., 1992).
This paper focuses generally on the use of contingent valuation to measure
people's values for environmental resources, rather than specifically on
natural resource damages. It will describe how researchers go about
conducting reliable surveys. It then addresses some common objections to
surveys and, lastly, considers the compatibility between contingent
valuation and economic theory.
Conducting Reliable Surveys
In all research, details matter. How a contingent valuation survey is
conducted is crucial. While there is no panacea, various procedures have
been developed in recent years that enhance the credibility of a survey and
make it more likely to produce reliable results. These touch all aspects,
including sampling, instrument development, formulation of the valuation
scenario, questionnaire structure, and data analysis. The main ways of
assuring reliability are summarized here.
Suppose one approached people in a shopping mall, made them put their bags
down for a moment, and asked them what was the most they would be willing to
pay for a sea otter in Alaska or an expanse of wilderness in Montana. This
is how the President of American Petroleum Institute and other critics have
characterized contingent valuation (DiBona, 1992). The essence of their
argument is summarized in titles such as "Ask a Silly Question" and "Pick a
Number" (Anon., 1991; Bate, 1994). It does not require any unusual
perspicacity to see that this approach is unlikely to produce reliable
results. For precisely this reason, it is not what good contingent valuation
researchers do, and it is not what was recommended by the NOAA Panel on
Contingent Valuation (Arrow et al., 1993) described in Portney's paper in
this issue.
Serious surveys of the general public avoid convenience sampling, such as
stopping people in the street; they employ statistically based probability
sampling.[2] They also avoid self-administered surveys, such as mail surveys
or questionnaires handed out in a mall, because of the lack of control over
the interview process. For a major study, the NOAA Panel recommended
in-person interviews for their superior reliability. Furthermore, interviews
should occur in a setting that permits respondents to reflect and give a
considered opinion, such as their home. Unless the study deals with consumer
products, shopping malls are a poor choice. Indeed, the only contingent
valuation study where people were stopped for a few minutes in a mall was
one performed for Exxon (Desvousges et al., 1992).
The crux is how one elicits value. The two key developments have been to
confront subjects with a specific and realistic situation rather than an
abstraction, and to use a closed-ended question which frames the valuation
as voting in a referendum.
A common temptation is to characterize the object of valuation in rather
general terms: "What would you pay for environmental safety?" "What would
you pay for wilderness?" The problem is that these are abstractions.
People's preferences are not measured in the abstract but in terms of
specific items. "Paying for wilderness" is meaningless; what is meaningful
is paying higher taxes or prices to finance particular actions by somebody
to protect a particular wilderness in some particular manner. Therefore, one
wants to confront respondents with something concrete. Moreover, one should
try to avoid using counter factuals. "What would you pay not to have had the
Exxon Valdez oil spill?" is utterly hypothetical because one cannot undo the
past. By contrast, "What would you pay for this new program that will limit
damage from any future oil spills in Prince William Sound?" offers something
that is tangible.
The goal in designing a contingent valuation survey is to formulate it
around a specific commodity that captures what one seeks to value, yet is
plausible and meaningful. The scenario for providing the commodity may be
real; if not, the key is to make it seem real to respondents. They are not
actually making a payment during the interview, but they are expressing
their intention to pay. The vaguer and less specific the commodity and
payment mechanism, the more likely respondents are to treat the valuation as
symbolic. To make the payment plausible, one needs to specify the details
and tie them to provision of the commodity so this cannot occur without
payment. There should be a clear sense of commitment; for example, if the
program is approved, firms will raise prices, or the government taxes, so
there is no avoiding payment once a decision is made.[3]
Until the mid-1980s, most contingent valuation surveys used some version of
an open-ended question, like "What is the most you would be willing to pay
for . . . ?" Since then, most major contingent valuation studies have used
closed-ended questions like "If it cost $x, would you be willing to pay this
amount?" or "If it cost $x, would you vote for this?" Different people are
confronted with different dollar amounts. Plotting the proportion of "yes"
responses against the dollar amount traces out the cumulative distribution
function of willingness-to-pay.[4]
Of course, if people carried utility functions engraved in their brains, the
question format would not matter. But they don't, and it does matter. In
this country, posted prices are the norm rather than bargaining. In market
transactions people usually face discrete choices: here is an item, it costs
$x, will you take it? Similarly in voting. Moreover, there is abundant
evidence that respondents find the open-ended willingness-to-pay question
much more difficult to answer than the closed-ended one; for market and
nonmarket goods alike, people can generally tell you whether they would pay
some particular amount, but they find it much harder to know what is the
most that they would possibly pay. Indeed, the experience with open-ended
willingness-to-pay questions for market goods is that people are more likely
to tell you what the good costs than what it is worth to them. In addition
to being less realistic and harder to answer, the open-ended format creates
incentives which are different from those in the closed-ended format. With
the open-ended format, as with an oral auction, there are strategic reasons
for stating less than one's full value--a theoretical result strongly
supported by experimental evidence. This is not so with a closed-ended
format; there, the NOAA Panel held, there is no strategic reason for the
respondent to do other than answer truthfully.[5]
For these reasons, the NOAA Panel considered the closed-ended format
combined, where possible, with a voting context the most desirable for
contingent valuation: "The simplest way to approach the valuation problem,"
it held, "is to consider a contingent valuation survey as essentially a
self-contained referendum in which respondents vote to tax themselves for a
particular purpose" (p. 20). This is a rather different conception of
contingent valuation from asking silly questions of passers-by.
In his introduction to this symposium, Portney describes other ways to make
a contingent valuation questionnaire more reliable: providing adequate and
accurate information; making the survey balanced and impartial; insulating
it from any general dislike of big business; reminding respondents of the
availability of substitutes, and of their budget constraint; facilitating
"don't know" responses; allowing respondents to reconsider at the end of the
interview. Several steps can be taken to eliminate any perception of
interviewer pressure. At the outset, the interviewer can assure respondents
that there are no "right" answers. Before asking the voting question, to
legitimate a negative response, the interviewer could say something like:
"We have found that some people vote for the program and others vote
against. Both have good reasons for voting that way," and then list some
reasons for saying "no."[6] Another possibility is if the interviewer does
not actually see the respondents' votes, for example by having them write on
a ballot placed in a sealed box.
A recent innovation, considered essential by the NOAA Panel, is a
"debriefing" section at the end of the survey. This checks respondents'
understanding and acceptance of key parts of the contingent valuation
scenario. For example, was the damage as bad as described? Did you think the
program would work? Did you think you really would have to pay higher taxes
if the program went through? This also probes the motives for their answer
to the willingness-to-pay question. What was it about the program that made
you decide to vote for it? Why did you vote no? Moreover, throughout the
survey, all spontaneous remarks by the respondent are recorded verbatim as
they occur. After the survey, the interviewer is debriefed and asked about
the circumstances of the interview, how attentive the respondent was,
whether the respondent seemed to understand the questions and appeared
confident in his responses. In this way, one creates a rich portrait of the
interview. This information can be exploited in the data analysis. One can
monitor for the misunderstandings, measure statistically how they affected
respondents' willingness-to-pay, and adjust accordingly. For example, if a
subject who voted "yes" appeared to be valuing something different than the
survey intended, this case can be dropped or the "yes" converted to a "no."
With any data, different statistical procedures can produce different
results. The closed-ended format raises several statistical issue, for
example, one might summarize the willingness-to-pay distribution by using
its mean, or its median, or another quantile. The mean is extremely
sensitive to the right tail of the distribution; that is, to the responses
of the higher bidders. For this reason, if the mean is to be used, a
nonparametric or bounded influence approach is highly recommended for
fitting the willingness-to-pay distribution. The median, by contrast, is
usually very robust (Hanemann, 1984). Another issue is that the choice of
dollar bids affects the precision with which the parameters of the
willingness-to-pay distribution are estimated; significant improvements can
be achieved by using optimal experimental designs (Kanninen, 1993).
Statistical techniques can also be used to probe for yea-saying or other
response effects, and correct for them if they are present (Hanemann and
Kanninen, forthcoming).
While none of these alone is decisive, taken together they are likely to
produce a reliable measure of value. Apart from the expense of in-person
interviews, they are all eminently feasible.[7] Other essential ingredients
are relentless attention to detail and rigorous testing of the instrument,
usually in collaboration with survey experts, so that the researcher
understands exactly how it works in the field and is sure it communicates
what was intended.
It is no coincidence that the handful of studies that Diamond and Hausman
select from the contingent valuation literature in their companion paper in
this issue violate most of these precepts, as do the Exxon surveys reported
in Hausman (1993). None uses in-person interviews. Many are
self-administered. Most use open-ended questions. None is cast as voting.[8]
Many ask questions with a remarkable lack of detail.[9] Several seem
designed to highlight the symbolic aspects of valuation at the expense of
substance.[10] The Exxon surveys were designed and fielded in great haste,
with little pretesting, just at a time when federal agencies were gearing up
for natural resource damage regulations.[11] The only way to justify this is
to make the tacit assumption that, if contingent valuation is valid, details
of its implementation should not matter. This is fundamentally wrong:
measurement results are not invariant with respect to measurement practice
in any science.
Objections to Surveys
McCloskey (1985, p. 181) observes that economists generally dislike surveys:
"Economists are so impressed by the confusions that might possibly arise
from questionnaires that they have turned away from them entirely, and
prefer the confusions resulting from external observation." In this section,
I discuss four common objections to surveys.
Surveys are Vulnerable to Response Effects
Small changes in question wording or order sometimes cause significant
changes in survey responses (Schuman and Presser, 1981). Since virtually all
data used in economics come from surveys (including experiments, which are a
form of survey), and all surveys are vulnerable to response effects, it is
important to understand why these arise and how they can be controlled. A
consensus is beginning to emerge based on insights from psychology and
linguistics. Answering survey questions requires some effort, usually for no
apparent reward. Respondents must interpret the meaning of the question,
search their memory for pertinent information, integrate this into a
judgment, and communicate the judgment to the interviewer. Although some are
motivated to make the effort, others may become impatient, disinterested, or
tired. Instead of searching for an accurate and comprehensive answer, they
satisfice, just aiming for some response that will be accepted. Furthermore,
interviews are interactions governed by social and linguistic norms that
shape assumptions and expectations. Viewing respondents as satisficing
agents following norms of conversation has proved helpful in interpreting
survey data, explaining response effects, and designing more effective
surveys (Groves, 1989; Krosnick, 1991).
Not all response phenomena are equally intractable. Some, such as order
effects (for example, bias towards the first item in a list), can be
detected and controlled, either by choosing the sequence that produces a
conservative result or by randomizing the order of items across interviews.
A second type of effect is where there is a shift in meaning. This is
substance, not noise. For example, similar words turn out to mean different
things: "allow" is not the same as "not forbid," nor "higher prices" the
same as "higher taxes."[12] Or there are framing effects, where subjects
respond differently to situations the researcher saw as equivalent. It has
been shown through debriefings that the subjects perceived the situations as
substantively different, because either the researcher induced an unintended
change in meaning or context, or the subjects made inferences that went
beyond the information given (Frisch, 1993).[13] In each case, the shift in
meaning is a source of error only if the researcher is unaware of it.
Through rigorous testing with cognitive techniques, the researcher can come
to understand exactly what the instrument means to people, and what they
mean in response.[14]
A third phenomenon arises from the inherent difficulty of the task assigned
the respondent. In recalling past events or behavior, for example,
respondents resort to rounding, telescoping (time compression) and other
inferential strategies that yield inaccurate reports of magnitudes and
frequencies.[15] Bradburn et al. (1987) emphasize that factual and
attitudinal surveys share many similar cognitive processes and errors. There
is no easy solution for recall errors. This continues to be a problem for
many data used by economists,[16] though not for contingent valuation data
since there is no recall.
One cannot avoid the fact that surveys, like all communication, are
sensitive to nuance and context and are bound by constraints of human
cognition. One tries to detect discrepancies and repair them, but they
cannot be entirely ruled out. It is important to keep a sense of proportion.
As far as I know, nobody has stopped using data from the Current Population
Survey, Consumer Expenditure Survey, Monthly Labor Survey, or Panel Study on
Income Dynamics because there are response effects in such surveys. The same
should apply to contingent valuation surveys.
The Survey Process Creates the Values
It has been asserted that contingent valuation respondents have no real
value for the item, but just make one up during the course of the interview:
the process creates the values that it seeks to measure. Debriefings can
identify whether subjects were inattentive or unfocused and offered hasty or
ill-considered responses, and these can be discarded if desired. But, the
issue raised here is more fundamental. Diamond and Hausman feel they know
real preferences when they see them, and they do not see them in contingent
valuation. Based on the debriefing statements in Schkade and Payne (1993)
that show most subjects, faced with an open-ended willingness-to-pay
question, think about either what the item could cost or what they have
spent on something remotely similar, Diamond and Hausman conclude that these
people are just making up their answer rather than evincing "true economic
preferences." But, what are "true economic preferences"? If a subject
responds thoughtfully to a question about voting to raise taxes for a public
good, by what criterion is that not a valid preference?
It is true that economists often assume consumer choice reflects an
individual's global evaluation of alternatives, a "top-down" or
"stored-rule" decision process. The stored-rule notion traces back to Hobbes
and the English empiricists who conceived of cognition in terms of storing
and retrieving "slightly faded copies of sensory experiences" (Neisser,
1967). Wilson and Hodges (1992) call this the "filing cabinet" concept of
the mind. It long dominated not only economics but also psychology. But it
is now being abandoned in the face of accumulating evidence from the
neurosciences (Rose, 1992) and elsewhere that all cognition is a
constructive process--people construct their memories, their attitudes, and
their judgments. The manner of construction varies with the person, the
item, and the context. A general principle is that people are cognitive
misers: they tend to resolve problems of reasoning and choice in the
simplest way possible. This is the emerging consensus not only in survey
research, but also in social psychology, political psychology, and market
research (Martin and Tesser, 1992; Sniderman, Brody and Tetlock, 1991;
Payne, Bettman and Johnson, 1988).
For non-habituated and complex consumer choices, people often make
"bottom-up" decisions; that is, they make up a decision rule at the moment
they need to use it (Bettman, 1988). Olshavsky and Granbois (1979, p. 98)
found that "for many purchases a decision process never exists, not even on
first purchase." Bettman and Zins (1977) found that grocery shoppers
construct a choice heuristic "on the spot" about 25 percent of the time;
bottom-up construction of preferences occurred especially for meat and
produce "as might be expected, since consumers cannot really rely on brand
name for most choices of this type," less often for beverages and dairy
products "where either strong taste preferences may exist or only a limited
number of brands are available" (p. 81). This calls to mind a remark by
Robert Solow that the debriefings in Schkade and Payne "sound an awful lot
like Bob Solow in the grocery store." I suppose critics of contingent
valuation would consider that Solow does not have true economic preferences,
or that he has true economic preferences when buying milk but not meat.
The real issue is not whether preferences are a construct but whether they
are a stable construct. While this surely varies with circumstances, the
evidence for contingent valuation is quite strong. There is now a number of
test-retest studies in the contingent valuation literature, and these show
both consistency in value over time and a high correlation at the individual
level (Carson et al., 1994b). These levels of consistency are comparable to
the most stable social attitudes such as political party identification.
Ordinary People are Ill-Trained For Valuing the Environment
If, as the NOAA Panel suggests, the goal of a contingent valuation survey is
to elicit people's preferences as if they were voting in a referendum, then
prior experience or training are irrelevant. These are not a criterion for
voting.[17] Nor is their absence an argument against contingent valuation
per se. Through direct questioning, one can readily identify which
respondents knew of the issue before the interview, or before the oil spill,
and determine whether they hold different values from those who did not. How
one proceeds in calculating aggregate willingness-to-pay is something that
can be decided separately from the survey. Who has standing, and whose
values should count, are questions that we as economists have no special
competence to judge.
Survey Responses Can't Be Verified
There are three ways to validate contingent valuation results: replication,
comparison with estimates from other sources, and comparison with actual
behavior where this is possible. Replication is useful even on a small scale
both to see if results hold up and to check whether the instrument is
communicating as intended. This is the single best way for a researcher to
determine whether somebody's survey instrument works as claimed.
When contingent valuation measures direct use values, it may be possible to
make a comparison with estimates obtained through indirect methods. Knetsch
and Davis (1966) conducted the first test, comparing contingent valuation
and travel demand estimates (a method described in Portney's paper) of
willingness-to-pay for recreation in the Maine woods. The difference was
less than 3 percent. There are now over 80 studies, offering several hundred
comparisons between contingent valuation and indirect methods. The results
are often fairly close; overall, the contingent valuation estimates are
slightly lower than the revealed preference estimates and highly correlated
with them (Carson et al., 1994a).
The ideal is direct testing of contingent valuation predictions against
actual behavior. There are about ten such tests in the literature. Diamond
and Hausman mention only five of these. The ones not mentioned yield results
quite favorable to contingent valuation.
Bohm (1972) conducted the first test, where subjects in Stockholm were asked
their willingness-to-pay to see a new TV program. In five treatments, the
program was shown if the group raised 500 Kr, with actual payment based in
various ways on stated willingness-to-pay. A sixth treatment asked subjects
what was the highest amount they would have given if they had been asked to
pay an individual admission fee. The mean response was 10.2 Kr (about $2)
when the group was asked a hypothetical question, versus an overall average
of 8.1 Kr when the group actually paid. The difference between contingent
valuation and non-contingent valuation means was not statistically
significant in four of the five cases.
Bishop and Heberlein (1990) conducted a series of experiments with hunters
who had applied for a deer-hunting permit in a favored game preserve run by
the state of Wisconsin. The most relevant for current practice is an
experiment in which they wrote to two groups of hunters offering to sell
them a permit at a specified price. In one case, this was a real offer; in
the other, it was asked as a hypothetical question. Estimated
willingness-to-pay was $31 in the real sale versus $35 in the hypothetical
sale, a statistically insignificant difference.
Dickie, Fisher and Gerking (1987) offered boxes of strawberries door-to-door
at different prices. One treatment was a real offer--the household could buy
any number of boxes at this price. The other asked how many boxes they would
buy if these were offered at the given price. The resulting two demand
curves were not significantly different. The parameter estimates were
actually more robust over alternative model specifications for the
hypothetical than the actual data (Smith, 1994).
Carson, Hanemann and Mitchell (1986) tested the accuracy of voting
intentions in a water quality bond election in California in 1985.
Closed-ended contingent valuation questions were placed on the Field
California Poll a month before the vote, using different figures for the
household cost. Adjusted for "don't know" responses, the predicted
proportion of yes votes at the actual cost was 70-75 percent. The ballot
vote in favor was 73 percent.
Cummings, Harrison and Rutstrom (1993) offered subjects small commodities at
various prices. For one group, it was a real sale. A second group was first
asked a hypothetical contingent valuation question--this item is not
actually for sale but, if it were, would you buy it now? The experimenter
then announced that, after all, she would sell the item, but they should
feel free to revise their answer. When juicers were the item, 11 percent
actually bought them in the real sale; with the second treatment, 41 percent
said they would buy it if it were on sale, but then only 16 percent did. The
41 percent and 11 percent are significantly different. With calculators, 21
percent would buy in the hypothetical sale, versus 8 percent in the real
sale. One wonders whether some respondents interpreted the question as "if
you needed a juicer, would you buy this one?" Smith (1994) shows that the
calculator responses do not generate a downward sloping demand curve for
either the actual or hypothetical data. The experimental procedure contained
nothing to emphasize commitment or counteract yea-saying in the hypothetical
treatment. Cummings and his colleagues have recently added wording like the
"reasons to say no" mentioned earlier. In one case, this reduced the
hypothetical yes for calculators from 21 percent to 10 percent, not
significantly different from the real 8 percent; in another there was no
effect (Cummings, 1994).
Other contingent valuation tests have used open-ended payment questions,
with predictable difficulties. Boyce et al. (1989) measured
willingness-to-pay and willingness-to-accept for a house plant, with mixed
results; Neill et al. (1994) measured willingness-to-pay for a map and a
picture, with negative results. Both confound the issue by comparing
contingent valuation responses to an experimental auction, begging the
question of whether auction behavior understates willingness-to-pay.
Duffield and Patterson (1991) and Seip and Strand (1992) compare actual and
hypothetical contributions to an environmental cause. Diamond and Hausman
focus on these studies because they showed a significant difference. But,
soliciting an intention to make a charitable donation is a poor test of
contingent valuation, because it invites less commitment than soliciting an
intention to vote for higher taxes. To make things worse, Seip and Strand
used members of the environmental group as the interviewers in their
hypothetical treatment, thus increasing pressures for compliance. They
compared hypothetical phone responses with responses to an actual mail
solicitation. Duffield and Patterson compared hypothetical mail
solicitations from the University of Montana with actual mail solicitations
from the Nature Conservancy. In both studies, the difference in survey
administration introduces a confounding factor which undermines the
comparison. [18]
A cleaner test is provided by Sinden (1988) who conducted a series of 17
parallel experiments soliciting actual and hypothetical monetary donations
to a fund for assisting soil conservation or controlling eucalypt dieback.
In all 17 cases, there was no statistical difference between actual and
hypothetical willingness-to-pay.
Thus, there is some substantial evidence for the validity of contingent
valuation survey responses, although more studies are certainly needed. Many
existing studies do not incorporate the refinements in contingent valuation
method, described earlier, that emphasize realism and commitment. In this
respect, the test by Carson, Hanemann and Mitchell (1986) points in the
right direction because it deals directly with expression of voting
intentions. The positive results in that study are consistent with other
evidence showing that polls in this country reliably indicate public
sentiment at the time they are taken, and polls close to an election are
generally accurate predictors of the outcome. [19] Kelley and Mirer (1974)
found voting intentions correctly predicted the actual vote in four
presidential elections for 83 percent of those respondents who voted.[20]
Surveys of purchase intentions in market research may not be accurate
predictors of subsequent purchase behavior, but surveys of voting intentions
are.[21]
Contingent Valuation and Economic Theory
Critics of contingent valuation like Diamond and Hausman, and their
coauthors in Hausman (1993), reject contingent valuation as a method of
economic valuation because the results of contingent valuation studies are
inconsistent with economic theory as they see it. These assertions have
become quite widely known. However, careful examination shows that in some
cases the claims are not supported by the findings in the contingent
valuation literature, and in others they rest on unusual notions about what
economic theory does or does not prescribe. I briefly review these issues
here, leaving a more detailed treatment to Hanemann (1994a).
Diamond and Hausman, and Milgrom (1993), make a number of statements about
what is a permissible argument in a utility function. They argue that people
should care about outcomes, not about the process whereby these are
generated. People should not care whether animals are killed by man or die
naturally. They should not care about details of provision or payment for a
commodity, only price. Above all, they should value things for purely
selfish motives. In their accompanying piece, Diamond and Hausman phrase
this argument by saying that respondents should not contemplate "what they
think is good for the country," because that reflects "warm glow" rather
than "true economic preferences."[22] From this perspective, contingent
valuation is unacceptable because it picks up existence values; for those to
be allowed in a benefit-cost analysis, Milgrom (1993, p. 431) argues, "it
would be necessary for people's individual existence values to reflect only
their own personal economic motives and not altruistic motives, or sense of
duty, or moral obligation."[23]
This criticism hardly comports with the standard view in economics that
decisions about what people value should be left up to them. For example,
Kenneth Arrow (1963, p. 17) wrote: "It need not be assumed here that an
individual's attitude toward different social states is determined
exclusively by the commodity bundles which accrue to his lot under each. The
individual may order all social states by standards he deems relevant." Or
as Gary Becker (1993, p. 386) writes: "[I]ndividuals maximize welfare as
they conceive it, whether they be selfish, altruistic, loyal, spiteful, or
masochistic." When estimating demand functions for fish prior to Vatican II,
no economist ever proposed removing Catholics because they were eating fish
out of a sense of duty. Nor, when estimating collective choice models, do we
exclude childless couples who vote for school bonds because they lack a
personal economic motive.
A more substantive matter is how willingness-to-pay varies with factors that
could reasonably be expected to influence it. This has been raised in
connection with the embedding effect and the income elasticity of
willingness-to-pay. Regarding the latter, Diamond and Hausman assert in this
issue that the income effects measured in typical contingent valuation
surveys are lower than would be expected if true preferences are measured.
McFadden and Leonard (1993, p. 185) make the more specific claim that an
income elasticity of willingness-to-pay less than unity constitutes grounds
for doubting the validity of the contingent valuation method. There is no
basis for either assertion. In the literature on the demand for state and
local government services in the United States, the income elasticities
generally fall in the range 0.3 to 0.6 (Cutler, Elmendorf and Zeckhauser,
1993). With charitable giving by individuals, the income elasticities
generally fall in the range of 0.4 to 0.8. (Clotfelter, 1985). The income
elasticities in the contingent valuation literature vary with the item being
valued, but are generally in this same range (Kristrom and Riera, 1994).
The term "embedding effect," introduced by Kahneman and Knetsch (1992), has
come to mean several different things. The general notion is captured in the
(mis)conception that, with contingent valuation, you get the same
willingness-to-pay if you value one lake, two lakes, or ten lakes.[24] This
combines three distinct notions. One assertion, which arises when the object
of preference is thought to be simply the number of lakes, is that
willingness-to-pay varies inadequately with changes in the scale or scope of
the item being valued. This is a scope effect. Alternatively, if each lake
is seen as a separate argument in the utility function, then the assertion
is that a given lake has quite different value if it is first, second or
tenth in a set of items to be valued--it gets a high value when the first,
but it adds little or nothing to total value when second or tenth. This is a
sequencing effect. Thirdly, with either preference structure, the
willingness-to-pay for a composite change in a group of public goods may be
less than the sum of the willingnesses-to-pay for the individual changes
separately. This is a sub-adaptivity effect.
The question of how willingness-to-pay varies with the scale or scope of the
item being valued in a contingent valuation survey has long been considered,
starting with Cicchetti and Smith (1973)who elicited hiker's values for
trips in a Montana wilderness area and found that the willingness-to-pay for
trips where other hikers were encountered on two nights was 34 percent lower
than the willingness-to-pay for trips with no encounters. Many other studies
have since reported comparable findings using both internal (within-subject)
and external (split-sample) scope tests, including meta-analyses by Walsh,
Johnson and McKean (1992) covering over 100 contingent valuation studies of
outdoor recreation, and Smith and Osborne (1994) on 10 contingent valuation
studies of air quality. Carson (1994) reviews 27 papers with split-sample
tests of scope and finds a statistically significant effect of scope on
willingness-to-pay in 25 of them.
The two exceptions are Kahneman and Knetsch (1992) and Desvousges et al.
(1992). Critics of contingent valuation rely heavily on these two studies
when asserting the absence of scope effects in contingent valuation.[25]
Some of the problems with these two studies have already been noted,
including their failure to use a closed-ended voting format, the
after-the-fact provision of information in Kahneman and Knetsch's "top-down"
procedure, and he use of brief shopping mall intercepts by Desvousges et
al.[26] The latter elicited people's willingness-to-pay for preventing the
deaths of migratory waterfowl. Three separate versions of the questionnaire
said that 2,000, 20,000, and 200,000 out of 85 million birds die each year
from exposure to waste-oil holding ponds that could be sealed under a new
program. Respondents were told that the deaths amounted to much less than 1
percent of the bird population, to less than 1 percent, and to about 2
percent. If respondents focused on the relative impact on the population, it
is hard to believe that they would have perceived any real difference among
these percentages. The results of the scope test depend crucially on how
much one trims the data to remove what are clearly outliers. With a 10
percent trim, one obtains a highly significant scope effect.[27] At any
rate, even if one regards these two studies as highly credible evidence that
respondents were insensitive to scope, they certainly do not represent the
majority finding in the contingent valuation literature regarding the
variation of willingness-to-pay with scope.
How much should willingness-to-pay vary with scope? Diamond (1993) asserts
that economic theory requires it to increase more than proportionately with
the number of bird deaths. The variables in his model are the number of
birds originally in the population, q[sub O], the number at risk of dying,
q[sub R], and the number of those that are saved, q[sub S]. Let q[sub F] is
equivalent or equal to q[sub O] - q[sub R] + q[sub s]. Diamond assumes that
people should care only about q[sub F], the ultimate number of birds, not
how many were alive initially, at risk, or saved. He also assumes
preferences are quasiconcave in q[sub O]. The two assumptions together imply
quasiconvexity in q[sub R], which is what makes the elasticity of
willingness-to-pay with respect to q[sub R] greater than unity. The
conclusion depends critically on the assumption of perfect substitution
between q[sub O], q[sub S] and -q[sub R]. When contingent valuation data
disconfirm this, Diamond dismisses the method. Others might be more inclined
to believe the data and drop the assumption.[28]
With regard to sequencing and sub-adaptivity effects, these effects are
certainly present in contingent valuation responses, but one expects them to
occur, and they can be explained in terms of substitution effects and
diminishing marginal rates of substitution. When the quality of one lake
improves, you value an improvement in a second lake less if the lakes are
what Madden (1991) calls R-substitutes, and more if they are R-complements.
Far from being inconsistent with economic preferences (Diamond et al., 1993,
pp. 48-49), sub-adaptivity is likely to be the norm: while all goods cannot
be R-complements, Madden shows they can all be R-substitutes.[29] Similarly,
R-substitution explains sequence effects: if the lakes are R-substitutes,
the willingness-to-pay for an improvement in one lake is lower when it comes
at the end of a sequence of changes in lake improvements than at the
beginning while the willingness-to-accept for the change in the lake is
higher when it comes later in a sequence (Carson, Flores and Hanemann,
1992).[30] It should come as no surprise that the value of one commodity
changes when the quantity of another varies: in other words, that
willingness-to-pay depends on economic context.[31]
For many economists, the ultimate argument against contingent valuation is
that it violates the habitual commitment of the profession to revealed
preference. Three points should be noted. First, one must distinguish
between private market goods and public goods. Revealed preference is harder
to apply to the latter, especially when they are national rather than local
public goods (Cropper, 1994). Second, revealed preference is not foolproof,
either. It involves an extrapolation from observation of particular choices
to general conclusions about preference. One relies on various auxiliary
assumptions to rule out factors that might invalidate the extrapolation.
Those assumptions are not themselves verifiable if one is restricted to
observed behavior. This can sometimes make revealed preference a relatively
hypothetical undertaking.[32] Third, there is no reason why observing
people's behavior and asking them about behavioral intentions and motives
should be mutually exclusive. Fathoming human behavior is never easy; one
should utilize every possible source of information.
Above all, one should take a balanced view of the difficulties with each
approach. As Sen (1973, p. 258) wrote, "we have been too prone, on the one
hand, to overstate the difficulties of introspection and communication and,
on the other, to underestimate the problems of studying preferences revealed
by observed behavior." In the debate on contingent valuation, critics have
shown a tendency to employ simplistic dichotomies. Surveys of attitudes are
fallible and subject to the vagaries of context and interpretation; surveys
of behavior are unerring. In the market place, people are well informed,
deliberate, and rational. Outside it, they are ignorant, confused, and
illogical. As consumers, people can be taken seriously; as voters, they
cannot. In particular instances, these assertions may be correct. As
generalizations, however, they are a caricature.
Conclusions
When cost-benefit analysis started in the United States in the 1930s,
economic valuation was generally perceived in terms of market prices. To
value something, one ascertained an appropriate market price, adjusted for
market imperfections if necessary, and then used this to multiply some
quantity. Two things changed this. The first was the recognition, prompted
by the "new welfare economics" of the 1940s and especially Hotelling's paper
on public utility pricing, that the appropriate welfare criterion is
maximization of aggregate consumers' plus producers' surplus. While market
prices can safely be used to value marginal changes for market commodities,
the impact of non-marginal changes is measured by the change in areas under
demand and supply curves. The second development was Samuelson's theory of
public goods and his finding that their valuation must be based on vertical
aggregation of individual demand curves.
Together, these developments led to an important paradigm shift--one that
contributed directly to the emergence of nonmarket valuation and is still
evident in the current debate on contingent valuation.[33] This shift
changed the focus of valuation away from market prices towards demand and
supply functions as the underlying repositories of value. These functions
are behavioral relations, and the implication of the paradigm shift was that
economics is not just the study of markets, but more generally the study of
human preferences and behavior.
The conceptual link to nonmarket valuation is the recognition that, while a
demand curve is not observable if there is no market for a commodity, there
still exists a latent demand curve that perhaps can be teased out through
other means. Indirect methods are one approach to doing this, and contingent
valuation is another. In both cases, the details of implementation have a
large impact on the quality of the results.
Faced with the assertion that contingent valuation surveys can never be a
reliable source of information either for benefit cost analysis or for
damage assessment, the NOAA Panel rejected this as unwarranted. Two years
later, there is now even more evidence from recent studies and literature
analyses to support the Panel's conclusion. However, it would be misleading
for me to suggest that contingent valuation surveys can be made to work well
in all circumstances. I am sure situations could exist where a contingent
valuation researcher might be unable to devise a plausible scenario for the
item of interest. Nor would I wish to argue that all contingent valuation
surveys are of high quality. The method, though simple in its directness, is
in fact difficult to implement without falling into various types of design
problems that require effort, skill and imagination to resolve. Each
particular study needs to be scrutinized carefully. But the same is true of
any empirical study.
While I believe in the feasibility of using contingent valuation to measure
people's value for the environment, I do not mean to advocate a narrow
benefit-cost analysis for all environmental policy decisions, nor to suggest
that everything can or should be quantified. There will be cases where the
information is inadequate, the uncertainties too great, or the consequences
too profound or too complex to be reduced to a single number. I am well
aware of the fallacy of misplaced precision. But this cuts both ways. It
also applies to those who suggest that it is better not to measure nonuse
values at all than to measure them through contingent valuation. I reply to
such critics by quoting Douglass North: "The price you pay for precision is
an inability to deal with real-world issues" (Wall Street Journal, 7/29/94).
Is expert judgment an alternative to contingent valuation? Experts clearly
play the leading role in determining the physical injuries to the
environment and in assessing the costs of clean-up and restoration.
Assessing what things are worth is different. How the experts know the value
that the public places on an uninjured environment, without resort to
measurement involving some sort of survey, is unclear. When that public
valuation is the object of measurement, a well-designed contingent valuation
survey is one way of consulting the relevant experts--the public ITSELF.
* I want to thank Richard Carson, Jon Krosnick, Robert Mitchell, Stanley
Presser and Kerry Smith for their helpful comments, and Nicholas Flores and
Sandra Hoffmann for excellent assistance. I also thank the editors, without
whom this paper would be far longer.
1 For a formal definition, see Hanemann (1994a).
2 DiBona's scenario actually was the practice in the 1930s when most surveys
were "brief encounters" on the street or in stores (Smith, 1987). The 1940s
saw the adoption of probability sampling, standardized survey techniques,
longer and more complex survey instruments, and in-depth focused interviews
(Merton and Kendall, 1946).
3 To underscore this, the interviewer may tell respondents that the
government uses surveys like this to find out whether taxpayers are willing
to pay for new programs it is considering.
4 The methodology here is to assume a random utility model for individual
preferences. This can be estimated using standard techniques for binary
choices. Bishop and Heberlein (1979) were the first to use this format; the
link with utility theory was developed in Hanemann (1984).
5 With auctions, it is well documented that formal matters and that oral
auctions generate lower prices than posted-price auctions. Why the surprise
when the same holds true for open versus closed-ended payment questions?
6 For example, the interviewer might note that some people prefer to spend
the money on other social or environmental problems instead, or they find
the cost is more than they can afford or than the program is worth, or they
cannot support the program because it would benefit only one area (Carson et
al., 1992).
7 Is there an acceptable alternative to in-person surveys? The NOAA Panel
felt mail surveys have significant problems rendering them unsuitable.
Telephone surveys avoid these problems, but preclude the use of visual aids
and need to be short. The most promising alternative is a mail/telephone
combination in which an information package is mailed to respondents who are
then interviewed by phone (Hanemann, Loomis and Kanninen, 1991). This
permits an extensive phone interview which seems to provide many of the
benefits of an in-person survey at much lower cost.
8 Two studies Diamond and Hausman cite as showing a lack of commitment in
contingent valuation, Seip and Strand (1992) and Duffield and Patterson
(1991), used open-ended questions about payment to an environmental charity.
Most of Seip and Strand's subjects who were followed up afterwards said that
they had been expressing their willingness-to-pay for environmental problems
generally, rather than the particular environmental group. Careful
pretesting would have discovered this beforehand.
9 This is notably a problem in Diamond et al. (1993).
10 Including Kahneman and Ritov (1993), Kahneman and Knetsch (1992), and
Kemp and Maxwell (1993). The last two employ a "top-down" procedure in which
respondents are given details of the item only after they value it. They are
first confronted with something broad, like "preparedness for disasters."
After stating their willingness-to-pay for the broad category, they are told
what it comprises and asked their willingness-to-pay for one of those
components. Then, they are told what this comprises, and so on. The change
in the quantity of any item is never specified.
11 Hanemann (1994a, b) critiques these studies.
12 And different words can mean the same thing, as in the movie Annie Hall
where Woody Allen and Diane Keaton are asked by their psychiatrists how
often they have sex. He says: "Hardly ever, maybe three times a week." She
says: "Constantly, I'd say three times a week." With consumer expenditure
surveys, Miller and Guin (1990) attest that life imitates art.
13 When there is incomplete information in a survey, respondents may go
ahead and make their own assumptions. Consequently, the researcher loses
control over his instrument. Diamond et al. (1993) is a contingent valuation
example.
14 On testing by federal survey agencies, see Tanur (1992). Lack of adequate
testing can explain some notable violations of procedural
invariance--respondents saw cues or meaning which the researcher didn't
intend and failed to detect. An example is the base rate fallacy where "when
no specific information was given, prior probabilities are properly
utilized; when worthless evidence is given prior probabilities are ignored"
(Tversky and Kahneman, 1974). A norm of conversation is to present
information one believes relevant. That this was the expectation of subjects
could have been detected through debriefings. On violations of
conversational norms in base-rate experiments, see Krosnick, Li and Lehman
(1990).
15 Some pronounced telescoping errors are to be found in the Alaska
recreation survey conducted by Hausman, Leonard and McFadden (1993).
16 Juster and Stafford (1991) and Mathiowetz and Duncan (1988) discuss
biases in labor supply estimates due to problems with bunching and
misreporting in Current Population Survey data. Atkinson and Micklewright
(1983) discuss errors in Family Expenditure Survey reports of income and its
components. Other inconsistencies between micro- and macro-data sets for the
household sector are discussed in Maki and Nishiyama (1995).
17 Voter ignorance is a constant refrain for Diamond and Hausman. They use
it to form a syllogism: voters are ill-informed, contingent valuation is
like a referendum, therefore contingent valuation respondents are
ill-informed. Both parts are false. Contingent valuation researchers take
pains to ensure their samples are representative and their questionnaires
intelligible, informative, and impartial, thus avoiding the vagaries of
turnout and biased advertising in election campaigns. This is why political
scientists are becoming interested in "deliberative polling"--in effect,
extended contingent valuation surveys (Fishkin, 1991). Many analysts see a
substantial core of rationality in voter behavior. Cronin (1989) finds
Magleby's (1984) assessment of voter ignorance in referenda overblown.
Fiorina (1981) and McKelvey and Ordeshook (1986) emphasize how campaign
protagonists use signals to inform voters. Lupia (1993) analyzes the
insurance reform battle in the 1988 California ballot and finds that
informational "short cuts" enabled poorly informed voters to act as though
they were well informed. What Sniderman (1993) calls "the new look in public
opinion research" stresses how ordinary citizens use the information at hand
to make sense of politics.
18 The problem with mail surveys is that people may think the survey is junk
mail and throw it out unopened. Duffield and Patterson made no allowance for
the difference in sponsor identity on the envelope, which could explain the
difference in response rates (Schuman, 1992). Response rates apart, the
pattern of contributions was similar in the two treatments. Seip and Strand
made no allowance for the fact that phone and mail solicitations generally
have different response rates. Infosino (1986) found a sales rate three
times higher with telephone than mail in an AT&T marketing effort.
19 Diamond and Hausman seem troubled that voters change their minds during
the course of an election campaign. They cite a 1976 electricity rate
proposition in Massachusetts where support went from 71 percent in February
to 25 percent in the November ballot. They fail to mention the reasons.
Magleby (1984, p. 147) identifies opposition spending as the chief cause of
such opinion reversals, and that certainly occurred in 1976--opponents
outspent supporters more than threefold. In May, the Dukakis administration
came out against it, as eventually did businesses, the unions, hospitals,
colleges, and major newspapers.
20 Ajzen and Fishbein (1980) offer some reasons to expect a high level of
attitude-behavior correspondence for voting in terms of their theory of
reasoned action.
21 One reason for the difference is timing: unlike elections, people
generally control the timing of their market purchases. The result is they
may end up buying the commodity, but later than they said (Juster, 1964).
This is especially likely for durables, the focus of much literature, since
their durability permits delay in replacement. This is consistent with
findings that purchase intentions are significantly more accurate for
nondurables than durables (Ferber and Piskie, 1965); intentions not to
purchase durables are highly accurate (Theil and Kosobud, 1968); and
predictions of the brand selected when the purchase does occur tend to be
highly accurate (Ajzen and Fishbein, 1980; Warshaw, 1980).
22 "Warm glow" is simply a red herring. I have seen no empirical evidence
that people get a warm glow from voting to raise their own taxes, whether in
real life or in a contingent valuation study.
23 Milgrom (1993) also asserts that using contingent valuation to measure
altruistic preferences creates double counting. His analysis has three
flaws. First, it depends on the particular specification of the utility
function, as Johansson (1992) notes; if the argument of the utility function
is another's consumption rather than his utility, there is no double
counting. Second, it derives its force from the auxiliary assumption that
the respondent does not realize that the other people for whom he cares will
have to pay, too; this is not a problem in a referendum format. Third, in
many contingent valuation studies the object of the altruism is often
wildlife--sea otters, for example. Since those creatures are not surveyed,
the issue of double counting is moot.
24 Though widely believed, this is a myth. It may be traced to Kahneman
(1986), which is usually cited as showing that respondents were willing to
pay the same amount to clean up fishing lakes in one region of Ontario as in
all of Ontario. His data actually show a 50 percent difference. Moreover,
the survey involved a brief telephone interview using an open-ended
willingness-to-pay question. It provided no detail on how and when the
cleanup would occur. Respondents may not have seen cleaning up all the lakes
as something likely to happen soon.
25 Also, in their contingent valuation survey, Diamond et al. (1993, pp.
45-46) mention that, using a Kruskal-Wallis test, they found no difference
in willingness-to-pay for three wilderness areas ranging in size from
700,000 to 1.3 million acres. If they had run a simple regression of
willingness-to-pay on acreage, they would have found a significant scope
effect.
26 Other questions about Kahneman and Knetsch are raised by Harrison (1992)
and Smith (1992).
27 How the survey was administered clearly affected the results. Schkade and
Payne (1993) used the same questionnaire as Desvouges et al., but slowed
respondents down and made them think about their answer. Their data show a
different pattern of willingness-to-pay responses, and a significant
relationship between willingness-to-pay and the percentage of birds killed
(Haneman, 1994b).
28 Some, while not sharing Diamond's extreme position on the elasticity of
willingness-to-pay, still hold that contingent valuation responses vary
inadequately with scale. People's perceptions undoubtedly differ from
objective measures of attributes. But this is not just a feature of
contingent valuation. In psychophysics, it has been known since the 1880s
that there is a general tendency for judgments of magnitude to vary
inadequately. Observers standing at a distance overestimate the height of
short posts, and underestimate that of tall ones; people reaching quickly
for an object overestimate small distances and angles, and underestimate
large ones; subjects matching loudness of a tone to a duration overestimate
the loudness of short tones, and underestimate the loudness of long ones;
people overestimate infrequent causes of death, and underestimate frequent
ones; small probabilities are overestimated, large ones underestimated
(Poulton, 1989). This "response contraction bias" in judgment or rating is
an authentic feature of how people perceive the world, not an artifact of
contingent valuation.
29 If the intention of the Diamond et al. (1993) contingent valuation survey
was to test the adding-up of willingness-to-pay, it was strangely designed
for the purpose. The survey stated that there were 57 federal wilderness
areas in the Rocky Mountain states, without identifying them, and said that
there now was a proposal to open these to commercial development. In one
version, respondents were told that seven unidentified areas had already
been earmarked for development, and were asked their willingness-to-pay to
protect an eighth area, identified as the Selway Bitterroot Wilderness. In
another, respondents were told that eight unnamed areas had been earmarked
for development and asked their willingness-to-pay to protect a ninth area,
identified as the Washakie Wilderness. In a third version, respondents were
told that seven unnamed areas had been earmarked for development and asked
their willingness-to-pay to protect two areas identified as Selway and
Washakie. In all three cases, respondents were not told the identity or fate
of the other 48 or 49 areas. Given that respondents were not indifferent
among wilderness areas, as evidenced by the regression mentioned in note 25,
I leave it to the reader to decide whether the surveys constitute a sensible
basis for testing the adding up of willingness-to-pay.
30 In natural resource damages, where willingness-to-accept is the relevant
welfare measure, this implies that the usual practice of taking the inured
resource as the first item in any possible valuation sequence is a
conservative procedure.
31 The practical implications are that, when one values a program, it be
placed in whatever sequence applies under the circumstances, and that one
take care when extrapolating results in a benefits transfer exercise because
the values might change with the difference in circumstances (Hoehn and
Randall, 1989).
32 Revealed preference estimates are sensitive to the measurement of price,
which is often uncertain and precarious for disaggregated commodities
(Pratt, Wise and Zeckhauser, 1979; Randall, 1994). The price at which demand
falls to zero, needed to estimate consumer's surplus, may lie outside the
range of the observed data and be estimated inaccurately (for example. one
knows travel cost only for participants, or one believes that participants
and nonparticipants have different preferences). This can cause revealed
preference to produce a less reliable estimate of use value than contingent
valuation (Hanemann, Chapman and Kanninen, 1993). With other variables there
may be inadequate variation in the data (for example, attributes are
correlated across brands). Hence, revealed preference data alone may yield a
less reliable estimate of demand functions than contingent valuation choice
data, and one may need to combine both types of data for best results
(Adamowicz, Louviere and Williams, 1994).
33 For an account of the development of nonmarket valuation generally, see
Hanemann (1992).
References
Adamowicz, W., J. Louviere, and M. Williams, "Combining
Revealed and Stated Preference Methods for Valuing
Environmental Amenities," Journal of Environmental
Economics and Management, 1994, 26, 271-92.
Ajzen, Icek, and Martin Fishbein, Understanding Attitudes and
Predicting Social Behavior. New Jersey: Prentice-Hall,
Inc., 1980.
Anonymous, "'Ask a Silly Question . . . ' Contingent Valuation
of Natural Resource Damages," Harvard Law Review, June
1992, 105, 1981-2000.
Arrow, Kenneth, J., Social Choice and Individual Values, 2nd
ed., New Haven: Yale University Press, 1963.
Arrow, Kenneth, et al., Report of the NOAA Panel on Contingent
Valuation, Washington, D.C.: January 1993, p. 41.
Atkinson, A. B., and J. Micklewright, "On the Reliability of
Income Data in the Family Expenditure Survey, 1970-1977,"
Journal of the Royal Statistical Society (A), 1983,
146(I), 33-53.
Bate, Roger, "Pick a Number: A Critique of Contingent
Valuation Methodology and Its Application in Public
Policy." Competitive Enterprise Institute, Environmental
Studies Program, Washington, D.C., January 1994.
Bateman, Ian, and Ken Willis (eds.), Valuing Environmental
Preferences: Theory and Practice of the Contingent
Valuation Method in the US, EC and Developing Countries.
Oxford, UK: Oxford University Press, forthcoming.
Becker, Gary S., "Nobel Lecture: The Economic Way of Looking
at Behavior," Journal of Political Economy, June 1993,
101(3), 385-409.
Bettman, James R., "Processes of Adaptivity in Decision
Making," Advances in Consumer Research, 1988, 15, 1-4.
Bettman, J. R., and M. A. Zins, "Constructive Processes in
Consumer Choice," Journal of Consumer Research, September
1977, 4, 75-85.
Bishop, Richard C., and Thomas A. Heberlein, "Measuring Values
of Extramarket Goods: Are Indirect Measures Biased?"
American Journal of Agricultural Economics, December
1979, 61, 926-30.
Bishop, Richard C., and Thomas A. Heberlein, "The Contingent
Valuation Method." In Johnson, Rebecca L., and Gary V.
Johnson, eds., Economic Valuation of Natural Resources:
Issues, Theory, and Applications, Boulder: West-view
Press, 1990, 81-104.
Bohm, Peter, "Estimating Demand for Public Goods: An
Experiment," European Economic Review, 1972, 3, 111-30.
Boyce, R. R., et al., "Experimental Evidence of Existence
Value in Payment and Compensation Contexts." Paper
presented at the USDA W-133 Annual Meeting, San Diego,
California, February 1989.
Boyle, Kevin J., Michael P. Welsh, and Richard C. Bishop, "The
Role of Question Order and Respondent Experience in
Contingent-Valuation Studies," Journal of Environmental
Economics and Management, 1993, 25, S-80-S-99.
Bradburn, Norman M., Lance J. Rips, and Steven K. Sheyell,
"Answering Autobiographical Questions: The Impact of
Memory and Inference on Surveys," Science, April 1987,
236, 157-161.
Briscoe, John, et al, "Toward Equitable and Sustainable Rural
Water Supplies: A Contingent Valuation Study in Brazil,"
World Bank Economic Review, May 1990, 4, 115-34.
Brookshire, David S., Mark A. Thayer, William D. Schulze, and
Ralph C. d'Arge, "Valuing Public Goods: A Comparison of
Survey and Hedonic Approaches," American Economic Review,
1982, 72, 165-77.
Carson, Richard T., "Contingent Valuation Surveys and Tests
of Insensitivity to Scope." Paper presented at the
International Conference on Determining the Value of
Non-marketed Goods: Economic Psychological, and Policy
Relevant Aspects of Contingent Valuation Methods, Bad
Hamburg, Germany, July 1994.
Carson, Richard T., and Nicholas E. Flores, "Another Look at
'Does Contingent Valuation Measure Preferences:
Experimental Evidence'--How Compelling is the Evidence?"
Economics Department, University of California, San
Diego, December 1993.
Carson, R., N. Flores, and W. M. Hanemann, "On the Creation
and Destruction of Public Goods: The Matter of
Sequencing," working paper 690, Agricultural and Resource
Economics, University of California, Berkeley, 1992.
Carson, Richard T., Nicholas E. Flores, Kerry Martin and
Jennifer Wright, "Contingent Valuation and Revealed
Preference Methodologies: Comparing the Estimates for
Quasi-Public Goods," Discussion Paper 94-07, University
of California, San Diego, May 1994a.
Carson, Richard T., W. Michael Hanemann, and Robert Cameron
Mitchell, "The Use of Simulated Political Markets to
Value Public Goods," Economics Department, University of
California, San Diego, October 1986.
Carson, Richard T., Kerry Martin, Jennifer Wright, "A Note on
the Evidence of the Temporal Reliability of Contingent
Valuation Estimates," working paper, University of
California, San Diego, Economics Department, July 1994b.
Carson, Richard T., and Robert Cameron Mitchell, "The Value
of Clean Water: The Public's Willingness to Pay for
Boatable, Fish-able, and Swimmable Quality Water," Water
Resources Research, 1993, 29, 2445-54.
Carson, R., et al., A Contingent Valuation Study of Lost
Passive Use Values Resulting from the Exxon Valdez Oil
Spill, Report to the Attorney General of Alaska, Natural
Resource Damage Assessment, Inc. La Jolla, CA, November
1992.
Carson, Richard T., et al., A Bibliography of Contingent
Valuation Studies and Papers, Natural Resource Damage
Assessment, Inc., La Jolla, CA, March 1994c.
Chase, S. B., ed., Problems in Public Expenditure Analysis.
Washington, D.C.: Brookings Institution, 1968.
Cicchetti, Charles J., and V. Kerry Smith, "Congestion,
Quality Deterioration, and Optimal Use: Wilderness
Recreation in the Spanish Peaks Primitive Area," Social
Science Research, 1973, 2, 15-30.
Ciriacy-Wantrup, S. V., "Capital Returns from
Soil-Conservation Practices," Journal of Farm Economics,
November 1947, 29, 1188-90.
Clotfelter, Charles T., Federal Tax Policy and Charitable
Giving. Chicago: The University of Chicago Press, 1985.
Cronin, Thomas E., Direct Democracy: The Politics of
Initiative, Referendum, and Recall. Cambridge: Harvard
University Press, 1989.
Cropper, Maureen L., "Comments on Estimating the Demand for
Public Goods: The Collective Choice and Contingent
Valuation Approaches." Paper presented at the DOE/EPA
Workshop on "Using Contingent Valuation to Measure
Non-Market Values," Herndon, VA, May 19-20, 1994.
Cummings, Ronald G., "Relating Stated and Revealed
Preferences: Challenges and Opportunities." Paper
presented at the DOE/EPA Workshop on "Using Contingent
Valuation to Measure Non-Market Values," Herndon, VA, May
19-20, 1994.
Cummings, Ronald G., David S. Brook-shire, and William D.
Schulze, et al., eds. Valuing Environmental Goods: An
Assessment of the Contingent Valuation Method. Totowa,
New Jersey: Rowman and Allanheld, 1986.
Cummings, Ronald G., Glenn W. Harrison, and E. E. Rustrom,
"Homegrown Values and Hypothetical Surveys: Is the
Dichotomous Choice Approach Incentive Compatible?"
Economics Working Paper Series, B-92-12, Division of
Research, College of Business Administration, The
University of South Carolina, February 1993.
Cutler, David, Douglas W. Elmendorf, and Richard J.
Zeckhauser, "Demographic Characteristics and the Public
Bundle," National Bureau of Economic Research, Cambridge,
NBER Working Paper No. 4283, February 1993.
Desvousges, William H., et al., Measuring Nonuse Damages Using
Contingent Valuation: An Experimental Evaluation of
Accuracy. North Carolina: Research Triangle Institute
Monograph, 1992.
Diamond, Peter A., Jerry Hausman, Gregory K. Leonard, and Mike
A. Denning, "Does Contingent Valuation Measure
Preferences? Experimental Evidence." In Hausman, J. A.,
ed., Contingent Valuation: A Critical Assessment. New
York: North-Holland, 1993, 41-89.
DiBona, Charles J., "Assessing Environmental Damage," Issues
in Science and Technology, Fall 1992, 8, 50-54.
Dickie, M. A. Fisher, and S. Gerking, "Market Transactions and
Hypothetical Demand Data: A Comparative Study," Journal
of American Statistical Association, March 1987, 82,
69-75.
Duffield, John W., and David A. Patterson, "Field Testing
Existence Values: An Instream Flow Trust Fund for Montana
Rivers." Presented at the American Economics Association
Annual Meeting, New Orleans, Louisiana, January 4, 1991.
Ferber, Robert, and Robert A. Piskie, "Subjective
Probabilities and Buying Intentions," Review of Economics
and Statistics, August 1965, 47, 322-25.
Fiorina, Morris P., Retrospective Voting in American National
Elections. New Haven: Yale University Press, 1981.
Fishkin, J. S., Democracy and Deliberation: New Directions for
Democratic Reform. New Haven: Yale University Press,
1991.
Freeman, A. Myrick, The Measurement of Environment and
Resource Values: Theory and Method. Washington, D.C.:
Resources for the Future, 1993.
Frisch, Deborah, "Reasons for Framing Effects," Organizational
Behavior and Human Decision Processes, 1993, 54, 399-429.
Groves, Robert M., Survey Errors and Survey Costs. New York:
John Wiley and Sons, 1989.
Hanemann, W. Michael, "Welfare Evaluations in Contingent
Valuation Experiments with Discrete Responses," American
Journal of Agricultural Economics, August 1984, 66,
332-41.
Hanemann, W. Michael, "Preface: Notes on the History of
Environmental Valuation in the USA." In Navrud, Stale,
ed., Pricing the Environment: The European Experience.
Oxford, UK: Oxford University Press, 1992.
Hanemann, W. Michael, "Contingent Valuation and Economics,"
Working Paper No. 697, Giannini Foundation of
Agricultural and Resource Economics, University of
California, Berkeley, February 1994a. To appear in
Willis, Ken, and John Corkindale, eds., Environmental
Valuation: Some New Perspectives, Walling-ford, Oxon, UK:
CAB International, forthcoming.
Hanemann, W. Michael, "Strictly For the Birds: A
Re-examination of the Exxon Tests of Scope in CV,"
working paper, Giannini Foundation of Agricultural and
Resource Economics, University of California, Berkeley,
August 1994b.
Hanemann, W. Michael, and B. J. Kanninen, "Statistical
Analysis of CV Data." In Bateman, I., and K. Willis, eds.
Valuing Environmental Preferences: Theory and Practice
of the Contingent Valuation Method in the US, EC and
Developing Countries. Oxford: Oxford University Press,
forthcoming.
Hanemann, W. Michael, David Chapman, and Barbara Kanninen,
"Non-Market Valuation Using Contingent Behavior: Model
Specification and Consistency Tests." Presented at the
American Economic Association Annual Meeting, Anaheim,
California, January 6, 1993.
Hanemann, W. M., J. Loomis, and B. Kanninen, "Statistical
Efficiency of Double-Bounded Dichotomous Choice
Contingent Valuation," American Journal of Agricultural
Economics, 1991, 73, 1255-63.
Harrison, Glenn W., "Valuing Public Goods with the Contingent
Valuation Method: A Critique of Kahneman and Knetsch,"
Journal of Environmental Economics and Management, 1992,
23, 248-57.
Hausman, J. A., ed., Contingent Valuation: A Critical
Assessment. New York: North-Holland, 1993.
Hoehn, J.P., and A. Randall, "Too Many Proposals Pass the
Benefit Cost Test," American Economic Review, June 1989,
79, 544-51.
Infosino, William J., "Forecasting New Product Sales from
Likelihood of Purchase Ratings," Marketing Science, Fall
1986, 5, 372-384.
Johansson, Per-Olov, "Altruism in Cost-Benefit Analysis,"
Environmental and Resource Economics, 1992, 2, 605-13.
Jones-Lee, Michael W., M. Hammerton, and P. R. Philips, "The
Value of Safety: Results of a National Sample Survey,"
Economic Journal, March 1985, 95, 49-72.
Juster, F. Thomas, Anticipations and Purchases: An Analysis
of Consumer Behavior. Princeton: Princeton University
Press, 1964.
Juster, F. Thomas, and Frank P. Stafford, "The Allocation of
Time: Empirical Findings, Behavioral Models, and Problems
of Measurement," Journal of Economic Literature, 1991,
29, 471-522.
Kahneman, Daniel, "Valuing Environmental Goods: An Assessment
of the Contingent Valuation Method: The Review Panel
Assessment." In Cummings, R. G., D. S. Brookshire, W. D.
Schulze, et al., eds., Valuing Environmental Goods: An
Assessment of the Contingent Valuation Method. Totowa,
New Jersey: Rowman & Allanheld, 1986, 185-94.
Kahneman, Daniel, and Jack L. Knetsch, "Valuing Public Goods:
The Purchase of Moral Satisfaction," Journal of
Environmental Economics and Management, 1992, 22, 57-70.
Kahneman, Daniel, and Ilana Ritov, "Determinants of Stated
Willingness to Pay for Public Goods: A Study in the
Headline Method," unpublished, Department of Psychology,
University of California, Berkeley, 1993.
Kanninen, B. J., "Optimal Experimental Design for
Double-Bounded Dichotomous Choice Contingent Valuation,"
Land Economics, May 1993, 69, 128-46.
Kelly, S., and T. W. Mirer, "The Simple Act of Voting,"
American Political Science Review, 1974, 68, 572-91.
Kemp, Michael A., and Christopher Maxwell, "Exploring a Budget
Context for Contingent Valuation Estimates." In Hausman,
J. A., ed., Contingent Valuation: A Critical Assessment.
New York: North-Holland, 1993, 217-69.
Knetsch, J. L., and R. K. Davis, "Comparisons of Methods for
Recreation Evaluation." In Kneese A. V., and S.C. Smith,
eds., Water Research, Baltimore: Resources for the Future
Inc., Johns Hopkins Press, 1966, 125-42.
Kristrom, Bengt, and Pere Riera, "Is the Income Elasticity of
Environmental Improvements Less Than One?" Paper
presented at the Second Conference on Environmental
Economics, Ulvong, Sweden, June 2-5, 1994.
Krosnick, Jon A., "Response Strategies for Coping with the:
Cognitive Demands of Attitude Measures in Surveys,"
Applied Cognitive Psychology, 1991, 5, 213-36.
Krosnick, Jon A., Fan Li, and Darrin R. Lehman,
"Conversational Conventions, Order of Information
Acquisition, and the Effect of Base Rates and
Individuating Information on Social Judgments," Journal
of Personality and Social Psychology, 1990, 59, 1140-52.
Krutilla, John V., "Conservation Reconsidered," American
Economic Review, September 1967, 57, 777-86.
Lupia, Arthur, "Short Cuts versus Encyclopedias: Information
and Voting Behavior in California Insurance Reform
Elections," working paper, Department of Political
Science, University of California, San Diego, April 1993.
Madden, Paul, "A Generalization of Hicksian q. Substitutes and
Complements with Application to Demand Rationing,"
Econometrica, September 1991, 59, 1497-1508.
Magleby, David B., Direct Legislation, Voting on Ballot
Propositions in the United States. Baltimore and London:
The John Hopkins University Press, 1984.
Maki, Atsushi, and Shigeru Nishiyama, "Consistency Between
Macro- and Micro-Data Sets in the Japanese Household
Sector," Review of Income and Wealth, 1993, 39, 195-207.
Martin, Leonard L, and Abraham Teaser, eds., The Construction
of Social Judgments. New Jersey: Lawrence Erlbaum
Associates, chapter 2, 1992, 37-65.
Mathiowetz, Nancy A., and Greg J. Duncan, "Out of Work, Out
of Mind: Response Errors in Retrospective Reports of
Unemployment," Journal of Business & Economic Statistics,
1988, 6, 221-29.
McCloskey, Donald, The Rhetoric of Economics. Madison: The
University of Wisconsin Press, 1985.
McFadden, Daniel, and Gregory K. Leonard, "Issues in the
Contingent Valuation of Environmental Goods:
Methodologies for Data Collection and Analysis." In
Hausman, J. A., ed., Contingent Valuation: A Critical
Assessment. New York: North-Holland, 1993, 165-215.
McKelvey, Richard D., and Peter C. Ordeshook, "Information,
Electoral Equilibria and the Democratic Ideal," Journal
of Politics, 1986, 48, 909-37.
Merton, Robert K., and Patricia L. Kendall, "The Focused
Interview," American Journal of Sociology, 1946, 51,
541-57.
Milgrom, Paul, "Is Sympathy an Economic Value? Philosophy,
Economics, and the Contingent Valuation Method." In
Hausman, J. A., ed., Contingent Valuation: A Critical
Assessment. New York: North-Holland, 1993, 417-41.
Miller, Leslie A., and Theodore Downes-Le Guin, "Reducing
Response Error in Consumers' Reports of Medical Expenses:
Application of Cognitive Theory to the Consumer
Expenditure Interview Survey," Advances in Consumer
Research, 1990, 17, 193-206.
Mitchell, Robert Cameron, and Richard T. Carson, Using Surveys
to Value Public Goods: The Contingent Valuation Method.
Washington, D.C.: Resources for the Future, 1989.
Navrud, Stale, Pricing the European Environment. New York:
Oxford University Press, 1992.
Neill, Helen R., et al., "Hypothetical Surveys and Real
Economic Commitments," Land Economics, May 1994, 70,
145-54.
Neisser, Urlic, Cognitive Psychology, Appleton-Century-Crofts,
Educational Division, New York: Meredith Corporation,
1967.
Oates, W., "Comments on Estimating the Demand for Public
Goods: The Collective Choice and Contingent Valuation
Approaches." Paper presented at the DOE/EPA Workshop on
Using Contingent Valuation to Measure Non-Market Values,
Hemdon, VA, May 19-20, 1994.
Olshavsky, Richard W., and Donald H. Granbois, "Consumer
Decision Making--Fact or Fiction?" Journal of Consumer
Research, September 1979, 6, 93-100.
Payne, J. W., J. R. Bettman, and E. J. Johnson, "Adaptive
Strategy Selection in Decision Making," Journal of
Experimental Psychology: Learning, Memory, and Cognition,
1988, 14, 534-52.
Poulton, E. C., Bias in Quantifying Judgments. Hove, UK:
Lawrence Erlbaum Associates, 1989.
Pratt, John W., David A. Wise, and Richard Zeckhauser, "Price
Differences in Almost Competitive Markets," Quarterly
Journal of Economics, May 1979, 93, 189-212.
Randall, Alan, "A Difficulty with the Travel Cost Method,"
Land Economics, February 1994, 70, 88-96.
Randall, Alan, Berry C. Ives, and Clyde Eastman, "Bidding
Games for Valuation of Aesthetic Environmental
Improvements," Journal of Environmental Economics and
Management, 1974, 1, 132-49.
Rose, Steven, The Making of Memory: From Molecules to Mind.
New York: Anchor Books, Doubleday, 1992.
Rosen, S., "Hedonic Prices and Implicit Markets: Product
Differentiation in Pure Competition," Journal of
Political Economy, January-February 1974, 82, 34-55.
Schelling, Thomas, "The Life You Save May Be Your Own." In
Chase, S., ed., Problems in Public Expenditure Analysis.
Washington, D.C.: Brookings Institution, 1968, 143-4.
Schkade, David A., and John W. Payne, "Where Do the Numbers
Come From? How People Respond to Contingent Valuation
Questions." In Hausman, J. A., ed., Contingent Valuation:
A Critical Assessment. New York: North-Holland, 1993,
271-303.
Schuman, H., remarks in transcript of Public Meeting of the
National Oceanic and Atmospheric Administration,
Contingent Valuation Panel, Washington, D.C.: NOAA,
Department of Commerce, August 12, 1992, p. 101.
Schuman, H., and S. Presser, Questions and Answers in Attitude
Surveys. New York: Academic Press, 1981.
Seip, K., and J. Strand, "Willingness to Pay for Environment
Goods in Norway: A Contingent Valuation Study with Real
Payment," Environmental and Resource Economics, 1992, 2,
91-106.
Sen, A. K., "Behavior and the Concept of Preference,"
Economica, August 1973, 40, 241-59.
Sinden, J. A., "Empirical Tests of Hypothetical Biases in
Consumers' Surplus Surveys," Australian Journal of
Agricultural Economics, 1988, 32, 98-112.
Smith, Tom W., "The Art of Asking Questions, 1936-1985,"
Public Opinion Quarterly, 1987, 51, 21-36.
Smith, V. Kerry, "Arbitrary Values, Good Causes, and Premature
Verdicts," Journal of Environmental Economics and
Management, 1992, 22, 71-89.
Smith, V. Kerry, "Lightning Rods, Dart Boards and Contingent
Valuation," Natural Resources Journal, forthcoming 1994.
Smith, V. Kerry, and William H. Desvousges, Measuring Water
Quality Benefits. Boston: Kluwer-Nijhoff Publishing,
1986.
Smith, V. Kerry, and Laura Osborne, "Do Contingent Valuation
Estimates Pass a 'Scope' Test?: A Preliminary Meta
Analysis." Presented at the American Economics
Association Annual Meeting, Boston MA, January 5, 1994.
Sniderman, Paul M., "The New Look in Public Opinion Research."
In Finifter, Ada W., ed., Political Science: The State
of the Discipline II. Washington, D.C.: The American
Political Science Association, 1993, 219-45.
Sniderman, Paul M., Richard A. Brody, and Phillip E. Tetlock,
Reasoning and Choice, Explorations in Political
Psychology. Cambridge: Cambridge University Press, 1991.
Tanur, Judith M., ed., Questions about Questions: Inquiries
into the Cognitive Bases of Surveys. New York: Russell
Sage Foundation, 1992.
Theil, Henri, and Richard F. Kosobud, "How Informative Are
Consumer Buying Intentions Surveys?" Review of Economics
and Statistics, February 1968, 50, 50-59.
Tversky, Amos, and Daniel Kahneman, "Judgment under
Uncertainty: Heuristics and Biases," Science, 1974, 185,
124-31.
Walsh, Richard G., Donn M. Johnson, and John R. McKean,
"Benefits Transfer of Outdoor Recreation Demand Studies:
19681988," Water Resources Research 1992, 28, 707-13.
Warshaw, Paul R., "Predicting Purchase and Other Behaviors
from General and Contextually Specific Intentions,"
Journal of Marketing Research, February 1980, 17, 26-33.
Wilson, Timothy D., and Sara D. Hodges, "Attitudes as
Temporary Constructions." In Martin, L., and A. Tesser,
eds., The Construction of Social Judgments. New Jersey:
Lawrence Erlbaum Associates, chapter 2, 1992, 37-65.
~~~~~~~~
By W. Michael Hanemann
W. Michael Hanemann is Associate Professor of Agricultural and Resource
Economics, University of California, Berkeley, California.
-------------------
Copyright of Journal of Economic Perspectives is the property of American
Economic Association and its content may not be copied or emailed to
multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email
articles for individual use.
Source: Journal of Economic Perspectives, Fall94, Vol. 8 Issue 4, p19, 25p.
Item Number: 9501060641
[Print] [Email] [Save] [Go To Full Text]