Jump to main content.

Can Contingent Valuation Measure Passive Use Values

Quick Links

Research Funding

Metadata


Contingent valuation (CV) is the only method currently available for practically measuring passive-use values. Because proposed laws may require that environmental regulations pass a benefit-cost test, CV has become central to the policy debate on environmental protection. Critics have raised substantial concerns about the ability of CV to produce reliable estimates of passive-use values. The most serious of these criticisms are: (1) hypothetical survey responses may not accurately predict actual behavior (validity), and (2) respondents may include alternative or additional sources of value when answering CV questions (embedding). A research team of economists and psychologists who span the range of viewpoints on CV has been formed to address this issue. Each researcher is, however, open to the possibility that CV may or may not be able to provide credible estimates of passive-use values. The group proposal structure will facilitate collaborative research. At the end of the project the group will attempt to provide a balanced assessment of the implications of the research.

To address the validity issue, the investigators propose to take advantage of a green pricing program offered by Niagara Mohawk Power Corporation. Niagara Mohawk plans to make annual offerings using an incentive compatible public good auction mechanism (a provision point with a money back guarantee) to provide specific green projects. This approach is unlike prior green pricing programs which have allowed free riding and have had poorly defined products. The first offering, if funded by customers, will provide an opportunity to purchase a landfill gas recovery project and plant 50,000 trees. Since Niagara Mohawk must offer the program to all customers, and since the air quality benefits of the landfill project will accrue locally, passive-use values may be present. This proposed research will utilize the second green offering, planned for the early summer of 1996. Five alternative CV elicitation methods, with two context variants, would be employed to attempt to estimate customer values prior to implementation of the second offering. These estimated values can then be compared with the actual customer values obtained in the public good auction. This opportunity to conduct a social experiment will help answer fundamental questions concerning the validity of CV, identify which methods, if any, are most appropriate, as well as allow for development of a calibration factor, etc. Niagara Mohawk has indicated its intention to support this proposal to NSF/EPA both by funding Cornell to evaluate the first offering, and by providing assistance in implementing the proposed NSF/EPA research.

To address the issue of embedding, a two part study is envisioned. An economic explanation for embedding has been developed which argues that, if environmental goods are close substitutes, one can explain embedding anomalies, i.e., if any two environmental goods are valued in order, and the first good always gets a much larger value than the second, no matter which good comes first, the result is plausible if all environmental goods are strong substitutes. As part of prior NSF and EPA sponsored research, the investigators have explored a multi-attribute choice (from among multiple options) approach which allows estimation of a random utility model of preferences for characteristics of environmental programs. This approach can identify substitutes in a choice environment (significant cross effects) if a sufficient sample size is employed (a Monte Carlo study of the required sample size has been completed). Thus, it is possible to test the proposed economic explanation for embedding . If the economic explanation fails, psychological explanations are the alternative. The next part of the proposed research will explore the psychology of embedding. The commodities to be used in this study are survival probabilities for endangered species, a topic of considerable importance in and of itself if reliable values can be obtained. Extensive pretesting will lead to a national survey using a split sample design with half of the respondents answering choice questions and half answering open ended willingness to pay questions (to allow exploration of order and other embedding effects).

Although the investigators hold differing opinions concerning the likely outcome of these experiments, the results of the proposed research will hopefully allow some consensus to be obtained. Those areas where disagreement remains will be clearly identified in the conclusions of the overall study.

Metadata

EPA/NSF ID:
R824688-010
Principal Investigators:
Schulze, William
Poe, Gregory
Kahneman, Daniel
Cameron, Trudy
Brown, Gardner
McClelland, Gary
Technical Liaison:
Research Organization:
Cornell University
Princeton University
California at Los Angeles, University of
Washington, University of
Colorado, University of
Funding Agency/Program:
EPA/ORD/Valuation
Grant Year:
1995
Project Period:
October 1, 1995 - September 30, 1997
Cost to Funding Agency:
$381,150
Project Status Reports:
For the year 1997

Objectives: Critics have raised substantial concerns about the ability of contingent valuation (CV) to produce reliable estimates of passive-use values. A serious criticism is that hypothetical survey responses may not accurately predict actual behavior (validity). The objective of our project is to address the validity issue by comparing alternative CV elicitation methods (open ended, payment card, dichotomous choice, multiple bounded discrete choice, and conjoint/stated preference) with actual participation in a Niagara Mohawk Power Corporation (NIMO) green pricing program that offered customers the opportunity to fund a landfill gas recovery project and plant 50,000 trees. The NIMO program used a provision point mechanism to address free riding.

Progress Summary:
Our research consisted of two efforts. First, we conducted laboratory experiments to explore the performance of single shot provision point mechanisms (SSPM) in large group (n>40) settings. Surprisingly, these mechanisms resulted in approximately demand revealing behavior. Provision point mechanisms have a Nash equilibrium where participants’ contributions sum to cost. Uncertainty in a single shot environment with many risk averse participants apparently motivates demand revelation. This is an important result both because a practical and reliable demand revealing public good mechanism is needed to calibrate CV methods, and because this mechanism can be used to facilitate actual funding of public goods by utilities, local communities, and environmental organizations.

The second part of our project consisted of a field test comparing actual and hypothetical participation levels in the NIMO Green Choice program. Two separate survey efforts were undertaken, as summarized in Table 1 on the following page. First, we conducted telephone interviews during which respondents were asked to sign up on the spot for the NIMO program (which had a $6 monthly cost). These responses can be compared to hypothetical dichotomous choice sign-ups for the program in a nearly identical phone interview at the single posted offer price of $6. We also asked an open ended willingness to pay (WTP) question in another treatment. Results from the phone survey are reported in the first three columns of Table 1. Appropriate statistical tests indicate that the hypothetical dichotomous choice sign-up rate was significantly different from the actual sign-up rate. The prediction from the open ended WTP question was much closer to, and not significantly different from, the actual level of participation. These results suggest that the open ended mechanism provides a more accurate estimate of predicted participation rates than the dichotomous choice method. However, any of the methods can be calibrated based on actual participation.

The second mail survey effort utilized the entire array of standard elicitation procedures used in CV surveys. Extremely preliminary results are shown in Table 1. Mail and phone survey results are similar both for dichotomous choice and open ended WTP. Surprisingly, the payment card approach produced the lowest estimate while the conjoint/stated preference approach produced the highest estimates of value and/or participation. The multiple bounded discrete choice approach is still being analyzed.

Accomplishments and Research Results: Our findings are significant both for advancing public good mechanism design and for calibrating CV methods. This is the first field study of a public good to simultaneously use a demand revealing mechanism for the actual good and to compare actual participation with predictions for the complete range of CV methods. Replication of these findings in other settings should be a priority for future research.

Phone Survey
Mail Surveys
Data for the mail survey are still being collected. Results are very preliminary since these data are incomplete and have not been verified or cleaned. At this point, only simple, trial statistical analyses have been conducted.
Actual $6 Sign-Ups
Hypothetical Open-Ended
Hypothetical Dichotomous Choice at $6
Hypothetical Open-Ended
Hypothetical Payment Card
Hypothetical Dichotomous Choice Var. $
Hypothetical Conjoint/
Stated Preference
Response Rate (%)
71.4
74.8
71.2
63.3
68.6
65.5
65.4
Final/
Current Sample (n)
142
294
259
266
285
717
325
Actual/
Estimated Sign-Up Rate at $6 per mo. for 12 mos. (%)
20.5
23.9
30.5
19.3
9.0
33.0
n.a.
Median WTP (Preliminary)
2.00
1.00
2.43
2.69
Table 1. Preliminary Results From NIMO Field Research.
Project Reports:
      Steven K. Rose, Jeremy Clark, Gregory L. Poe, Daniel Rondeau, and William D. Schulze, " The Private Provision of Public Goods: Tests of a Provision Point Mechanism for Funding Green Power Programs," Working Paper 97-02, Department of Agricultural, Resource, and Managerial Economics, Cornell University, July, 1997. This paper utilizes field and laboratory experiments to test the use of a provision point mechanism to finance renewable energy programs, commonly known as green pricing programs. In contrast to most green pricing programs, relatively high participation is found in the field, while laboratory results suggest that demand revelation is achieved by the mechanism in a single shot environment with a large group of potential participants.

      Daniel Rondeau, William D. Schulze, and Gregory L. Poe, "Voluntary Revelation of the Demand for Public Goods Using a Provision Point Mechanism," Working Paper 97-20, Department of Agricultural, Resource, and Managerial Economics, Cornell University, October, 1997; also forthcoming in Journal of Public Economics. A one-shot provision point mechanism with money-back guarantee and proportional rebate of excess contributions is tested in an induced value framework and in experimental environments chosen to mimic field conditions. The results show that this relatively simple mechanism is empirically demand revealing in the aggregate when used with large groups of students who have heterogeneous valuations for the public good. Approximately demand revealing behavior was obtained under three alternative information conditions. These results are an important step in the design of a mechanism simple enough to allow field applications, but capable of efficiently providing public goods through voluntary contributions.

      Gregory L. Poe, Jeremy Clark, and William Schulze, "Can Hypothetical Questions Predict Actual Participation in Public Programs? A Field Validity Test Using a Provision Point Mechanism," Working Paper 97-21, Department of Agricultural, Resource, and Managerial Economics, Cornell University, October, 1997; also Journal of Environmental Economics and Management – revise and resubmit status as of March, 1999.. Past field validity tests of contingent valuation have relied on voluntary contribution mechanisms to elicit actual willingness to pay, and are thus likely to overestimate hypothetical bias because of free riding in the actual contributions. This chapter argues that provision point mechanisms -- which have recently been shown to approximately reveal demand in large group public goods experiments -- should instead be used in contingent valuation validity testing, and employs such a mechanism in a validity study of green electricity pricing. Some upward hypothetical bias is found even when this improved mechanism is used. Calibration of hypothetical responses is also explored.

      Robert G. Ethier, Gregory L. Poe, and William Schulze, "A Comparison of Hypothetical Phone and Mail Contingent Vaualtion Rsponses for Green Pricing Electricity Programs," Working Paper 97-25, Department of Agricultural, Resource, and Managerial Economics, Cornell University, December, 1997; also forthcoming in Land Economics. This study provides the first contingent valuation phone-mail comparison that meets current standards for response rates, draws from a general population, is relevant to the valuation of general environmental goods, and allows comparisons with actual participation rates. Social desirability effects are found to be more prevalent in phone responses to subjective questions, but do not appear to affect hypothetical participation decisions: calibrated and uncalibrated hypothetical participation rates are statistically similar across modes. As such, neither mode appears to dominate from the perspective of providing more valid estimates of actual participation decisions.

      Trudy Ann Cameron, Gregory L. Poe, Robert G. Ethier, and William D. Schulze, Alternative Non-market Value-Elicitation Methods: Are the Underlying Preferences the Same? Submitted to Journal of Environmental Economics and Management. In a unique survey, six different random sub-samples of respondents were presented with an opportunity to value the identical environmental good, each via a different elicitation method. The methods include: an actual purchase decision at a single price, a referendum format with differing prices, an open-ended format, a payment card format, a multiple-bounded format, and a stated choice among an extended set of five alternatives (including the basic good plus three additional alternatives). We employ a common underlying indirect utility function and a stochastic structure that is also fully compatible across methods, allowing us to pool all of these different types of valuation data in one unified model. The different types of valuation data are entirely compatible with the same underlying set of homogeneous preferences, providing heteroscedastic errors across elicitation methods are permitted.

      Trudy Ann Cameron, Jeremy E. Clark, Robert G. Ethier, Gregory L. Poe, Daniel Rondeau, Steven K. Rose, and William D. Schulze, Can Contingent Valuation Measure Passive-Use Values? Final Report to the U.S. Environmental Protection Agency, March 31, 1999. Download full report here (740KB, Acrobat PDF file) FINAL_REPORT_R824688.PDF

Local Navigation


Jump to main content.