Stated Preferences,
Experiments &
Referenda
Shortcomings of revealed preference methods
1 Demand for many environmental goods cannot be estimated
Ex. – Existence value – if value is not connected with use, behavior is not
affected
Goods for which no logical market exists through
2 which value can be reflected
Ex – scenic vista/climate change
3 Serious econometric problems may arise while
estimating demand curve from the hedonic price
function
Constructed market Valuation tools
Valuation techniques that involve construction of a market
where one does not exist, to derive the demand curve
1 2
Stated Preference Experimental Markets
Consumers are directly asked what they The researcher constructs all the characteristics
would pay for an environmental good, if of the market, including trading money for a
there were a market. good and then observes behaviour.
Contingent Valuation 1 2 3
Hypothetical Valuation Laboratory experiments Field experiments Official referendum
Target population as subjects General population Voters are asked to
exchanging real money for in a realistic setting commit public money
experimental commodities for environmental
good
Stated Preferences
Finding an individual’s willingness to pay – by posing questions directly to the individual.
Imagine some situation that is typically outside the individual's experience and speculate on
how he/she would act in that situation.
Ex. Los Angeles (1980) – photographs depicting 3 different levels of air quality, people with poor air
quality – asked WTP to move to fair air quality
People with fair air quality - asked WTP to move to good air quality
Willingness to pay for cleaner drinking water in Seoul (Kwak & Rusell, 1994)
No water quality variation across neighborhoods – hedonic method – not possible.
Used a payment card with a range of reference expenditure amounts and people were asked to bid.
Developed individual’s Willingness to pay function, 𝑊𝑇𝑃 𝑞0, 𝑞1 = 𝑓(𝑃, 𝑞0 , 𝑞1 , 𝑄, 𝑌, 𝑇)
𝑞0, :Initial risk in water supply , 𝑞1 : risk after water supply has been made safe ,𝑃: price of market goods, 𝑄, qty. of other environmental goods,
𝑌 Income, 𝑇 assorted characteristics of the individual
Used this to construct an aggregate, population wide measure of WTP
To be able to support government regulatory actions, the CV study must be defensible and conducted
carefully, for which a fairly well-designed protocol is accepted.
Designing a contingent valuation study
Choosing
Design Market
Define Market Elicitation
Administration
Scenario Method
Experimental Estimation of
Sample Design Design WTPdfd
function
Defining Market Scenario
1 2 3 4 5 6
Information to If the The market Care must Educate the Respondents
be conveyed to respondents do scenario must be taken to respondent should be
the respondent not understand be rooted in avoid grating fully but reminded that
to place the the scenario, or real world on the prevent bias substitutes
respondent in
the right frame
2
think it is
implausible, they
experience,
including the
respondent’s
notion of
introduced by
incomplete or
exist.
Ex. If a beach is
of mind to give may substitute payment right or inaccurate contaminated,
meaningful an alternative vehicle (tax, fee, wrong information they should
scenario or not know that other
responses price)
beaches also
take it seriously
exist nearby.
Choosing Elicitation Method
1 2 3 4
Direct question The bidding game approach The payment card Referendum/Discrete Choice
Directly asking Start with WTP number and A card with a number of A WTP figure is offered to the
question but the seek a “yes” or “no” figures spanning the respondents, seeking a “yes”
problem is that since response. If respondent says range of responses which r “no” response only.
real markets do exist,
people are unfamiliar
2
“yes”, then gradually increase
the amount till a “no” is
can be expected.
- Cannot be used for
Different respondents are
offered different WTP
with the possible received. If respondent says telephone surveys figures. Most recommended
price rages and may “no”, gradually decrease the - Responses printed on approach since it minimises
give extreme amount till a “yes” is cards may serve as bias and requires a simple
responses increased. cues. vote.
- Suffers from starting point - However a large dataset is
bias. required to obtain
statistically significant results.
Design Market Administration
1 2 3 4
Mail Surveys Internet surveys Telephone surveys In-person surveys
Cheapest to administer but Becoming more Larger respondent base but Most expensive but most
Problems important by the - Expensive to administer reliable and therefore most
- Non-response (30% is day. One example is - Bias in terms of who recommended. But suffers
2
acceptable but if there is the internet panel answers the phone from two biases
a systematic reason for it survey where a (unemployed – more - Interviewer bias: difficult
-result questionable ) group of likely) for interviewer to always
- Survey must be simple respondents is - Junk telephone be neutral
for respondent to recruited to answer solicitations make it - Social desirability bias –
understand. online survey. difficult to reach respondents may trying
- The respondent may look potential respondents giving the WTP that they
through the entire feel is socially desirable.
survey. - Results obtained from
different interviewers may
not be comparable.
Sample Design
A random sample should be drawn from the sampling frame, for which we need
to precisely define the population from which the sample will be drawn.
Experimental Design
The typical goal of a CV survey is to develop statistically significant estimates for
WTP for a particular environmental good or to test a hypothesis about the WTP
for the hypothetical good. Considering the cost of data collection, the survey
should be constructed carefully (Design of the CV instrument and administration)
so that appropriate information could be collected in the most efficient manner,
without unintentional biases. The final and most important step in the
experimental design is the statistical analysis of the collected data.
Estimation of WTP function
The last step is to correctly estimate the WTP function. A good experimental
design must keep this aspect in mind before designing the CV instrument,
otherwise, some vital information might get skipped during survey .
Problems with Contingent Valuation
Controversial: Conflict of interest withing community. Many polluting firms would gain from
discrediting the approach.
Values elicited in CV surveys are not based on real resource decisions – they are hypothetical. Many
argue that without real resources at stake, the response to a WTP question is meaningless.
No budget constraint in a hypothetical survey – choices meaningless.
Ambiguity in what people are valuing/purchasing moral satisfaction by responding that you would
contribute substantially for better environmental quality.
Embedding: Inconsistencies on how people value individual environmental goods v/s a group of
environmental goods. This problem is even more when existence value is in question. It has been shown
in a study, that the WTP was the same for protecting a bird-specie, whether the number of birds were
2000, 20,000 or 200,000.
CV v/s Experimental Markets
Problem with CV
01 Hypothetical, no money at stake
Solution
02 Construct a market where none existed before
- Lab/ Field experiments
Purpose of experiments
03 Not valuation, but to test a theory in a controlled
setting
Valuation methods find distinction between
Valuation v/s Experiments
04 WTP and WTA but in experiments it is possible
to control for extraneous factors that may
drive a wedge between WTA and WTP
Laboratory Experiments
Involve a group of subjects who are paid a fixed amount to participate in the
survey for a few hours.
Participants make real decisions for real sources, exchanging real money for
experimental commodities.
Participants leave the experiment richer than when they began. (This induces
people to voluntarily participate.
Explore divergence between WTP and WTA as well as the validity of contingent
valuation.
Two famous experiments:
Sucrose Octa Acetate
Norfolk Pines
Demand for consuming foul tasting liquid - pollution (bad)
Results
Method
WTA starts very high when students
For WTP, students were asked to have little experience with SOA. After
Sucrose submit a bid to avoid tasting 1 repeated trials the WTA bids drop
Octa-Acetate oz. of SOA. Bids were arranged considerably, while WTP bids were
Foul-tasting liquid but
from highest to lowest. Fifth stable, changing only slightly as
non-toxic highest bid, was fixed as the auction trials repeated. => WTP is a
price. Higher bidders paid, while better measure than WTA,
lower bidders had to taste. particularly when the goods are
unfamiliar.
A Similar procedure was adopted
Performed with 8
for WTA, where each student Compared to the real values that
economics students
offered an amount they were emerged from experimental auctions,
accepting in compensation for the hypothetical measures of WTA
tasting the liquid. appeared to be biased upwards, while
WTP valuations were more consistent
with experimental values.
How much will you pay
to take this plant home?
How much will you
accept from me, if I want
to buy it from you?
“I will kill the plant.”
Use Value and Existence value of individual trees
Method Results
(statistically significant)
1. Each subject was given a plant
and was asked to submit a bid 1. The average WTA bid was 66% higher
Norfolk Island indicating a price at which they than the average WTP bid.
Pine tree would sell their plant to the
Commonly used as experimenter(WTA). A price was 2. The death threat raised both WTP (60%)
houseplants randomly generated and all and WTA (130%) => suggests the
offers below the price were importance of existence value over and
accepted. above use value.
3. Indicates importance of experimental
2. Subjects were asked to submit a markets in exploring questions of
bid indicating how much they valuation in case of environmental
Performed with 115 staff
members of a university, would pay to acquire such a tree. goods where markets rarely exist.
each paid $30-$40 for
3. Both experiments were repeated
participating in one
experiment only. after convincing the subjects that
if the plant remained with the
experimenter, he/she will kill the
tree
Field Experiments
Problem with Lab Ex.
01 Too contrived, and
students are not good representatives of the general population
Solution
02 Move out of the classroom into the real
world field experiments
Types of Field Ex.
03 Artefactual fields experiments – real people instead of university
students
( easiest )
Framed field experiments – general population but non-laboratory
setting
(most common)
Natural field experiments – subjects do not know they are part of
experiment
(most desirable)
Referenda
A voter referendum on the provisions, often accompanied
1 by a tax levy to pay for it. (Approach not very well
developed)
Observe how different groups of people are voting
2 and utilize the characteristics of the group to
explain the voting pattern and estimate the
demand curve.
3 Studies have shown that although it is difficult to
determine the price for environmental goods, since no
real markets exist for them, Referenda are able to
measure how income (and other characteristics) affect
demand.
Unit 5 done!!!
Any questions?
