You are here:
Updating Prior Methods for Non-Market Valuation: A Bayesian Approach to Combining Disparate Sources of Environmental Values
Despite the proliferation of contingent valuation and travel cost studies in recent years, little attention has been given to combining these two sources of data in estimating environmental values. Likewise, although there has been a notable increase in interest in the area of benefits transfer (applying the value of a studied resource to a unstudied resource), there remains room for additional rigor and more systematic methods in the implementation of these value transfers. In the research proposed here, a Bayesian framework will be used to provide a systematic approach to integrating and interpreting data from disparate sources. For both the data integration and benefits transfer problems, the Bayesian paradigm provides a natural and internally consistent way of framing the problem and for developing methodological solutions. The framework will be applied to the combination of contingent valuation and travel cost data, the combination of travel cost and contingent behavior data, and finally, to the transfer of travel cost or contingent valuation data from a set of studied sites to an unstudied policy site (i.e., for the purposes of benefits transfer). Specifically, the objectives of this proposed research are: (1) to develop and test Bayesian procedures for combining disparate sources of nonmarket valuations; (2) to develop and test Bayesian procedures for benefits transfer; and (3) to estimate the value of wetland restoration. This work will lay the foundation for applying Bayesian techniques to nonmarket valuation. In addition to methodological developments, this project will provide benefit estimates for an important and relatively little studied resource wetlands preservation in the Midwest. In recent years in Iowa, there has been a steady increase in interest in activities related to land use and the resource management on floodplains. Public programs such as the Wetland Reserve and Emergency Wetland Reserve, efforts of private conservation organizations, such as Ducks Unlimited, and the reality of continued flooding of the magnitude that occurred in 1993 have provided the motivation to investigate alternative land use and management of floodplains that were wetlands in their natural state. The valuation of wetlands in Iowa, both current and proposed, will provide much needed information to a variety of state and federal agencies, conservation groups, and policy analysts for the state legislature. In particular, one targeted group is the U.S. Department of Agriculture, Economics Research Service, members of whom have indicated their agency's need for values of wetland ecosystems in the Midwest. In addition, to the extent that the results of the wetlands valuation work can be transferred to other wetland sites in the United States, any agencies or groups with interests in the value of wetlands nationally would benefit from the results of this study. Furthermore, the methodology to employ the Bayesian paradigm to transfer these benefits would be readily available to policy analysts for the purpose of performing their own value transfers.
R825310-010Principal Investigators: Herriges, Joseph A.
Kling, Catherine L.Technical Liaison:Research Organization:
Iowa State UniversityFunding Agency/Program: EPA/ORD/ValuationGrant Year: 1996Project Period: October 1, 1996 - September 30, 1998Cost to Funding Agency: $210,199
- Project Reports
The three objectives of this research project were to: (1) develop and test a Bayesian procedure for combining disparate sources of nonmarket valuations, (2) develop and test Bayesian procedures for benefits transfer, and (3) estimate the value of wetland restoration in the state of Iowa. Although all three objectives have been met, the first two were altered, in that a classical paradigm was used rather than the Bayesian paradigm.
This project investigated the integration of disparate sources of data to improve the understanding of the value of nonmarket environmental goods, specifically wetlands in the state of Iowa. The integration of disparate data sources refers to the combination of contingent valuation and travel cost data, the combination of travel cost and contingent behavior data, and finally the transfer of travel cost or contingent valuation data from a set of studied sites to an unstudied policy site.
The first task to meeting these objectives was survey and sample design. The survey elicited several kinds of information from respondents. First, a series of questions was posed concerning the various visits these individuals made during the past year to wetland areas across the state. After establishing their current usage pattern, the survey respondents then were asked a contingent behavior question. In particular, the respondent was asked whether he/she would have taken at least one visit to a specific location if the price of visiting that location was higher. This question was followed with a question asking the respondent to indicate how many trips he/she would have taken to each of the wetlands areas assuming that this new, higher cost of taking the trip was in effect. A protest question completed this portion of the survey.
The next major section of the survey collected information on a variety of issues related to current knowledge about wetlands and opinions about how these areas should be managed. The third component of the survey contained a detailed scenario concerning one of two major wetland areas in the state. One of the scenarios concerned the Prairie Pothole Joint Venture, a program that has restored wetlands in several states in the upper midwest, including Iowa, as well as in portions of Canada. The second scenario was based on the Iowa River Corridor Project, a second major wetlands restoration effort in Iowa.
The final survey instrument was constructed and mailed to 6,000 Iowa households in February and March 1998, with an overall response rate of 58 percent. Using the software package PCMiler, travel time and travel costs to each aggregate site identified in the survey were constructed for each household and each destination in the 99 counties of Iowa.
Results on Combining Disparate Sources of Nonmarket Data. Economists began investigating the use of surveys to elicit consumers' willingness-to-pay for public goods over three decades ago. The vast majority of the early applications concerned environmental goods, although numerous other public goods have been the subject of valuation surveys. In the last decade, the use of surveys to elicit welfare values has come under debate by the profession at large. The fundamental question in this debate might be stated as follows: Can carefully designed survey methods provide informative data on consumers' willingness-to-pay for public goods, or does the hypothetical nature of these instruments render them irrelevant, regardless of how much attention is given to truth-revealing mechanisms in their construction?
Critics of SP have presented numerous arguments concerning likely sources of bias in the method. For example, it has been argued that survey respondents are unlikely to pay attention to their budget constraint in answering hypothetical questions. One procedure for assessing the validity of survey methods is to consider whether the preferences implied by answers to survey questions are consistent with preferences implied by observed behavior for the same individuals. In this approach, public goods for which there are market-like data or revealed preference (RP) data are valued using standard demand models or discrete choice approaches. Survey data also are collected and used to estimate the parameters of the same model as the RP data; tests of equality of the parameters are then undertaken to formally test consistency of the two data sources.
In this project, we formulated and tested specific hypotheses concerning sources of bias in survey data. The hypotheses were tested using revealed preference data to determine whether the underlying parameters are consistent between the two sources of data. Additionally, we formulated and tested hypotheses concerning sources of bias in revealed preference data. In this case, the hypotheses are tested using the stated preference data as the baseline from which to test consistency between the two data sources. In carrying out these sets of hypothesis tests, we carefully investigated alternative interpretations of the various results, noting particularly how the interpretation of the same results can vary with the "school of thought" of the reader.
Results on Procedures for Benefits Transfer. The Repeated Mixed Logit (RXL) model, a generalization of the standard Multinomial Logit Model, was estimated using the Iowa wetlands data. The RXL model allows the coefficients to vary randomly rather than being fixed. With this generalization, the model no longer exhibits the IIA property and can accurately represent a much wider variety of choice situations.
Using this model, the demand for wetlands in Iowa was estimated for each of five megazones (a grouping of the 15 identified wetlands areas into larger "sites"). The purpose was to examine the benefit transferability issue (i.e., whether the results from one megazone are transferable to megazones elsewhere). Recreators' willingness-to-pay for closing down all zones with each mezagone also was calculated.
The results indicated that cost, age, gender, and license holding all are significant determinants of individuals' choice decision in each megazone. Though some megazone models have similar coefficient estimates, individuals in different megazones have difference choice patterns, which is suggested by likelihood ratio tests. Also, the welfare measures associated with closing down all of the zones with each megazone vary across the megazones, which can be tested formally using nonparametric methods such as the Wilcoxon rank sum test. The results suggested that we should be cautious about transferring estimated results from one megazone to the other.
Results on Estimating the Value of Wetlands in Iowa. Using both parametric and semi-nonparametric estimators, a contingent valuation analysis of the Prairie Pothole Joint Venture suggests that households are, on average, willing to pay about $4 for the program to restore roughly 38,000 acres of wetlands. These valuations vary significantly by the characteristics of the households, with hunters and fishers willing to pay significantly more, and older Iowans willing to pay significantly less.
The average aggregate amount that Iowa households are willing to commit to wetland restorations that take place in the next 15 years is $150.14 per acre. When individuals responded to the contingent valuation question in our analysis, they were aware of the benefits and costs associated with wetlands. This includes the gain in flood control, water quality improvement, and wildlife habitat. The value we estimated then reflects both use and nonuse values for wetlands. Given the current land prices in Iowa, it is doubtful the value reported here could buy land already in a wetland state from private individuals for the purpose of maintaining the wetland indefinitely.
Cheng W. A repeated mixed logit model of Iowa wetlands recreation demand. Master's Thesis, Department of Economics, Iowa State University (to be submitted).
Azevedo C. Combining revealed and stated preference data in recreation demand modeling. Ph.D. Dissertation, Department of Economics, Iowa State University, 1999.
Wang X. Demand function estimation of recreation sites in wetlands of Iowa. Master's Thesis, Department of Economics, Iowa State University, 1999.
Crooker J. Valuing environmental amenities with nonparametric and semiparametric methods. Ph.D. Dissertation, Department of Economics, Iowa State University, 1998.
Herriges J, Phaneuf D. Choice set definition issues in a Kuhn-Tucker Model of recreation demand. Marine Resource Economics (accepted for publication).
Herriges J, Phaneuf D. Controlling for correlation across choice occasions and sites in a repeated mixed logit model of recreation demand. J Environ Econ Management (submitted for publication).
Crooker J, Kling C. Nonparametric bounds on welfare measures: a new tool for nonmarket valuation. J Environ Econ Management (accepted for publication). Also presented at the W-133 Regional Meeting, Tucson, AZ, 1999.
Herriges J, Kling C, Azevedo C. Ask a hypothetical question, get a valuable answer? Working Paper, 1999.
Herriges J, Kling C, Azevedo C. Linking revealed and stated preference data to test external validity. Presented at the American Agricultural Economics Association Meetings in Nashville, TN, 1999.
Crooker J, Kling C. Valuing Wetlands with Nonparametric Bounds. Presented at the First Annual Heartland Environmental and Resource Economics Workshop, Ames, IA, September 1999.
Supplemental Keywords: public policy, cost benefit, nonmarket valuation, contingent valuation, survey, willingness-to-pay, wetlands. , Economic, Social, & Behavioral Science Research Program, RFA, Scientific Discipline, Ecology and Ecosystems, Economics & Decision Making, Social Science, decision-making, Bayesian approach, aquatic ecosystems, benefits transfer, conservation, contingent valuation, cost benefit, cost effectiveness, cost/benefit analysis, decision analysis, economic incentives, ecosystem valuation, environmental assets, environmental policy, environmental values, midwest, non-market valuation, policy analysis, preference formation, psychological attitudes, public policy, public resources, public values, social psychology, standards of value, valuation, value transfers, valuing environmental quality, wetlands preservation
- Project Status Reports
- To develop and test Bayesian procedures for combining disparate sources of nonmarket valuations
- To develop and test Bayesian procedures for benefits transfer
- To estimate the value of wetland restoration
Progress Summary, Accomplishments, and Research Results:
As planned, we are well along in the process of designing and implementing the survey which provides the data required for this research. In particular, we have a completed questionnaire that has been revised based on the results of four focus groups we have held over the past three months. We have produced and mailed 600 copies of the survey as a pretest. Based on the results of the pretest we will revise the survey and are on schedule to mail approximately 6000 surveys to Iowa residents in February and March of 1998.
During the next year, we will complete the administration of both the pretest and the actual survey. The data will be coded and primary data analysis will begin. During the summer and fall of 1998, we anticipate initiating and completing the baseline valuation analysis discussed in our original proposal . We will also initiate work on the integration of travel cost and contingent valuation data, travel cost and contingent behavior data, and on the transfer of welfare measures across sites.
- From Proceedings of April 2-3, 1998 Workshop:
This project uses a Bayesian framework to provide a systematic approach to integrating and interpreting data from disparate sources. For data integration and benefits transfer problems, the Bayesian paradigm provides a natural and internally consistent way of framing the problem and for developing meth-odological solutions. The framework is being applied to the combination of contingent valuation and travel cost data, the combination of travel cost and contingent behavior data, and to the transfer of travel cost or contingent valuation data from a set of studied sites to an unstudied policy site. This project’s objectives are to: (1) develop and test Bayesian procedures for com- bining disparate sources of nonmarket valuations, (2) develop and test Bayesian procedures for benefits transfer, and (3) estimate the value of wetland restora- tion in the State of Iowa.
To explore the Bayesian methods in nonmarket valuation, the investigators proposed to undertake primary data collection via mail surveys. Specifically, 2,000 current fishing and hunting license holders in the State of Iowa are being randomly sampled as well as 4,000 Iowa residents drawn from the general popula- tion. A pretest of 600 Iowa residents was completed in the fall of 1997, and the full-blown study is being administered. The survey elicits several kinds of infor- mation from respondents. First, a series of questions is posed concerning the various visits that these individuals made during the past year to wetland areas across the state. For this task, the state is divided into 15 wetland regions. The number and location of the trips are elicited as well as information on the types of activities undertaken while at these sites (e.g., hunting, fishing, biking/hiking, nature viewing). After establishing their current usage pattern, the survey respondents are then asked a contingent behavior question. In particular, the respondent is asked whether he or she would have taken at least one visit to a specific location if the price of visiting that location was higher (this amount is the “bid”). Both the bid amount and the “location” are varied from survey to survey. The respondent is asked then how many trips he or she would have taken to each of the wetland areas, assuming that this new, higher cost of taking the trip was in effect. A protest question completes this portion of the survey.
The next major section of the survey collects information on a variety of issues related to current knowledge about wetlands and opinions about how these areas should be managed. This section is intended primarily to provide information for policy analysts in the state and for private and public agencies with interests in the amount of public awareness regarding wetland issues. The third component of the survey contains a detailed scenario concerning one of two major wetland areas in the state. One of the scenarios concerns the Prairie Pothole Joint Venture, a program that has restored wetlands in several states in the upper Midwest, including Iowa, as well as in portions of Canada. This restoration has been accomplished both by purchasing land outright from willing sellers and by developing a variety of easements where landowners retain the ownership of these lands, but agree to restore the land to its original prairie pothole wetland state. In this scenario, the Prairie Pothole Joint Venture is described, and a single-bounded contingent valuation question is asked concerning the respondent’s willingness to pay for increased conversions of land via this program.
The second scenario is based on the Iowa River Corridor Project, a second major wetlands restoration effort in the State of Iowa. The Iowa River Corridor is an area of saturated soils that experiences frequent flooding and encompasses approximately a 50-mile stretch along the Iowa River in central Iowa. As a consequence of the 1993 floods in the region, many landowners became interested in alternatives to traditional farming practices. As a result, the Natural Resource Conservation Service initiated the Iowa River Corridor Project, where landowners were given the option of enrolling their land in the Emergency Wetlands Reserve Program and would receive a one- time payment equal approximately to the value of their farm crops in exchange for a permanent easement on the land. The Natural Resource Conservation Service then restores the land to wetlands. In this version of the survey, the Iowa River Corridor Project is described to respondents. Respondents are asked about their willing- ness to pay for an additional 7,000 wetland acres in the region. Further, respondents are asked how many additional visits they would take to wetlands in the area.
The final section of the survey is common to both the Prairie Pothole and the Iowa River Corridor versions and gathers information concerning socioeconomic variables such as age, education levels, gender, and income. During the remainder of this project, efforts will turn towards the development and testing of Bayesian procedures for combining the various revealed and stated preferences obtained through the mail survey.
From previous status report:
Objectives of Research: