Contingent valuation method (CVM)

Outline
1. Overview and history
2. Structure of a CVM study
3. Criticisms
4. Lessons of experience with CVM in developing
countries

Readings
Perman et al., Ch. 14 (pp. 396-408)
Choe, Whittington, Lauria

1. Overview and history

Stated preference methods

Simplest method: contingent valuation
Involves a single good
Individuals are asked to state their maximum WTP or
minimum WTA for a change in the good
If individuals answer truthfully, their answers will
equal the Hicksian CS or ES for the change

More complex methods involve multiple goods, or

multiple attributes of single goods
Choice modeling (aka conjoint analysis)

CVM

Is survey-based
Often termed a direct valuation method, since
which individuals are asked to state directly their
WTP to obtain an environmental benefit or their
WTA to tolerate an environmental cost
Contingent: valuation is dependent on a
hypothetical scenario put to respondents

CVM: history
First applied in the U.S. in the 1960s
Came to prominence in early 1990s, due to use in
Exxon Valdez lawsuits
Number of studies:
1995: 2000 studies in 40 countries
2001: 5000+ studies in 100+ countries

Bilateral aid agencies and international

development banks are increasingly using CVM in
project appraisal and policy analysis

Principal advantages of CVM

1. In principle, willingness-to-pay (WTP) and
willingness-to-accept (WTA) responses elicited
by CVM equal theoretically correct (Hicksian)
monetary measures of utility changes

 Choice of WTP or WTA matters:

Hammack & Brown (1974): WTA = 4 WTP
Similar discrepancies in subsequent studies
WTP is preference of (most) applied economists:
Respondents often struggle to come up with bounded
responses to WTA questions, perhaps because more used to
paying for goods (subject to a budget constraint) than being
paid for them

Principal advantages of CVM

2. CVM can be used to estimate non-use (passive
use) values, like existence values

Often are public goods

Demand functions for public goods

The demand curve for a private good tells us:
If we have many consumers, what is the
maximum amount any single one of them will
pay if an additional unit is put up for sale?
The demand curve for a public good tells us:
If we have many individuals, what is the
maximum amount they will collectively pay if
an additional unit is put up for sale?
Implication for CVM: we can sum WTP responses
across respondents to estimate aggregate social value of
the environmental public good being valued

2. Structure of a CVM study

Example: Davao study

Households WTP for improved beach water

quality and improved sanitation
Two scenarios:
1. Water Quality Improvement Plan

Unspecified city-wide plan to clean up rivers and sea and

make Times Beach safe for swimming and other recreation

2. Sewer plus Treatment Plan

City-wide plan to construct sewer lines and treatment plants,

which will not only clean up rivers and sea but will also
enhance public health

Focus on scenario 1 here

Steps in a contingent valuation study

1. Define the good and the change in the good to be
valued

Water quality at Times Beach and other beaches near

Davao made safe again for swimming

2. Define the geographical scope of the market

General population of Davao: homeowners, renters,

squatters

3. Conduct focus groups on components of the

survey

Met with small groups of municipal officials and

household heads, to learn about recreation patterns
and sanitation situation

4. Pretest the survey instrument (questionnaire)

Based on focus group results, prepared draft

questionnaire and test-ran it on randomly selected
households

5. Administer the questionnaire to a stratified

random sample of households in Davao

1,200 households

6. Test for the reliability and validity of results

Analyzed whether responses were consistent with

demand theory

Components of questionnaire
1. Collect information on respondents past, present, and
expected future use of the good
2. Present a hypothetical scenario describing the change in
the good to be valued
3. Present the hypothetical payment mechanism and related
stipulations
4. Elicit the respondents WTP (bid elicitation procedure)
5. Collect information on respondents socioeconomic
characteristics, available substitutes and complements for
good being valued
6. Debrief respondent (e.g., check budget constraint) and
enumerator

See Davao survey instrument

3. Criticisms

Criticisms of CVM
1. Respondents fail to take CVM questions
seriously because they are non-binding
2. Respondents do not understand what they are
being asked to value
3. Respondents strategically manipulate the process
by distorting their true WTP
4. Respondents give answers that are inconsistent
with economic theory

Discrepancy in WTP/WTA was long thought to be

evidence of this

Potential biases in CVM studies

Operational bias: respondents understanding of the good might differ

from researchers
E.g., species interpreted as indicator for ecosystem: embedding or scope
effect

Hypothetical bias: respondents might ignore real-world costs and

benefits of consuming the good
E.g., budget constraint, crowding at a national park

Information bias: amount and type of information provided on

hypothetical good might affect stated WTP
But this is true of any consumption decision

Design bias
Starting-point bias: respondents might interpret starting point in bidding
game as conveying information about value of the good
Vehicle bias: choice of payment vehicle (e.g., entrance fee vs. higher taxes
to fund park) might affect stated WTP

Strategic bias: e.g., individuals misstate actual WTP (free-riding)

Investigating the reliability and

validity of CVM results
Design questionnaire to test for potential biases
(e.g., use different payment mechanisms)
Analyze whether bids are well-behaved from an
economic standpoint (e.g., higher for individuals
with higher income): estimate valuation
function
Replicate study
Compare CVM results to estimates from other
valuation methods
Ideally, compare bids to actual payments (e.g., fishing
licenses)

Current consensus on CVM

In wake of use of CVM in Exxon Valdez oil spill,
NOAA convened a blue-ribbon panel, chaired by
two Nobel laureates, to assess the method
Conclusion of the panel: CV studies can produce
estimates reliable enough to be the starting point of a
judicial process of damage assessment, including lost
passive use values (58 Federal Register 460, January
15, 1993)

But: many studies fail to comply with guidelines

recommended by the NOAA panel

Main concerns of NOAA panel

1.
2.
3.
4.
5.
6.

can produce results that appear to be inconsistent with assumptions

of rational choice (issue of scope)
responses sometimes seem implausibly large in view of the
many programs for which individuals might be asked to contribute
relatively few previous applications have reminded respondents
forcefully of the budget constraints under which all must operate
it is difficult to provide adequate information to respondents
and to be sure they have absorbed and accepted this information
in generating adequate estimates using the CV technique, it is
sometimes difficult determining the extent of the market
respondents in CV surveys may actually be expressing feelings
about public spiritedness or the warm glow of giving, rather than
actual willingness to pay for the program in question

Key NOAA panel recommendations

Randomly select samples of appropriate statistical size and
composition
Carefully pretest survey instrument
Conduct personal interviews (face-to-face is best)
Estimate WTP, not WTA
Design conservatively (downward bias in WTP)
Use referendum format to elicit bids, with no answer
option
Include follow-up questions to explain votes
Remind respondents about substitutes and budget
constraint

4. Lessons of experience with CVM

in developing countries

Whittington (1998)
Lessons learned from CVM studies in
developing countries
Main conclusion: although numerous issues
demand careful attention in CVM studies in
developing countries, in many respects conducting
high-quality CVM studies is easier in developing
countries than in industrialized countries

Explaining maximum WTP

Study in Haiti in late 1980s: What do you mean
the maximum I would be willing to pay? You
mean when someone has a gun to my head?
Is this what economists mean by WTP?

Im willing to pay, but not able

If this person willing to pay in an economic sense?

Interpreting yes/no responses

Interpretation of CVM questions can involve a
large cultural component
Hence, must word questionnaires carefully

During pretesting of questionnaire in Semarang,

Indonesia, all respondents said yes to
hypothetical connection fees and monthly tariffs
for improved water and sanitation services, no
matter how high
When debriefed enumerators, discovered that
respondents actually said, yes, but

Setting referendum prices

Upper bound of referendum prices should nearly
be the choke price: intercept of demand curve
But asking such a high price can make enumerators
look uninformed: everyone knows most people cant
afford such a high price

Lower bound should identify other end of demand

curve
But respondents might know that service cant be
provided so cheaply and thus doubt studys credibility
If agency funding the study is doing so to analyze
feasibility of prospective investment, it might not want
to give impression that it will charge a low price

 Very use of referendum method may spread

confusion about the problem being studied
1994 study in Mozambique: neighborhood leader
wanted to know why different households were being
asked to pay different prices for improved water service
Inferred that this might be actual policy, and it didnt
seem fair

Advantages of conducting CVM

studies in developing countries
Respondents are often quite receptive to listening
and considering questions posed
As Mozambique example indicates, they make consider
the study to be all too real
Malaysia study: respondents tried to pay enumerators

Response rates are typically very high

Helped by interviews being face to face

Survey costs are an order of magnitude lower

So, can have larger sample sizes, which permits more
split-sample testing of survey procedures