|Project Status Reports:|
For the Year 2002:
The objectives of this research project are to: (1) describe household choices in a framework that recognizes the assumptions and informational requirements of revealed preference methodologies (e.g., hedonic, random utility, and other models used to value changes in site specific amenities); and (2) incorporate a more explicit account of how economic activities impact environmental resources in these models. The specific focus of our research involves the quality of watersheds and related resources.
Our approach is intended to compare three methods: hedonic property value, random utility, and locational equilibrium models applied to Wake County, NC. To accomplish this comparison as a first step toward developing an integrated model, we proposed to use several existing databases, augment them with locationally delineated information on soil characteristics and water quality, and collect new data on a sample of households living in Wake County. The first database to be used for this analysis involves housing sales for Wake County, NC. The second involves measures of water quality likely to be impacted by land uses, relying on the U.S. Environmental Protection Agency (EPA), the U.S. Geological Survey (USGS), and the North Carolina Division of Water Quality (NC/DWQ) measurements. The last existing data set initially was proposed to be the 1994 EPA National Recreation Survey for NC. At this point, we have revised this focus to be the NC component of the 2000 National Recreation Survey.
Task 1: Select a Database for Hedonic Property Analysis and Integrate Housing With GIS Databases
We have access to two sets of information on housing sales. The first is from TransAmerica Intellitech (now merged with First American Real Estate Solutions). These data span a period from the early 1970s to the late 1990s. In the early years of the sample, the coverage of housing sales is not as complete as in most recent years. In addition, based on one of the investigator's (Ray Palmquist) collaborations with the Wake County Property Tax Office, we have all housing sales records from 1992 to 2000 for Wake County, and the assessment data for houses that did not sell in this period.
To date, we have developed three links to ARCVIEW shape files describing either watershed characteristics or neighborhoods. Two watershed definitions have been used for these connections. The first was developed by Palmquist and Phaneuf. Based on the USGS hydrologic unit classification (HUC), their analysis examined the segments of the Neuse River and its major tributaries with monitoring stations, and used the location of the stations to subdivide the HUCs into mini-HUCs so that each spatial unit has one water quality monitor.
The second geographic structure is derived from an initiative recently completed in Wake County. In November 2000, the Wake County Commissioners approved the development of a comprehensive watershed management plan. The plan was completed at the outset of our research, and it was made possible through persistent effort to obtain access to the GIS shape files used by the firm (CH2MHill) developing the plan, and to obtain the data the firm's analysts used in assessing the watersheds they defined for the county. Their analysis identified 81 watersheds in Wake County and assembled information on the variables.
The property sales records have been linked to both the Palmquist-Phaneuf classification scheme, allowing access to the water quality data linked to their mini-HUCs, and the CH2MHill classification scheme. Some of the features of the water quality measures being collected are described below.
These data also have been linked to shape files that describe spatial classifications of the county that could provide the basis for defining neighborhoods. The first of these is the census tracts for the county, the second is a 91-unit neighborhood structure developed by Randy Walsh that is based on the nodes that the Wake County Board of Education uses to route school buses.
These spatial delineations of the county have two roles. As we have already noted, they provide a basis for linking environmental quality measures to the individual properties falling within each spatial unit. In addition, in the case of the census tracts, they provide a basis for linking socioeconomic variables describing neighborhood characteristics.
Finally, they offer alternative definitions for the basic units to be used in defining the basic spatial units for the locational equilibrium models to be developed as part of this research.
Task 2: Design of Questionnaire for Mail Survey of Wake County Households
The new data to be collected in January and February 2003, are intended to supplement the housing sales price data. This effort is being designed to collect information about the economic and demographic characteristics of a sample of current homeowners in Wake County that could be matched to the housing data.
Our research proposal calls for a sample of approximately 5,000 homeowners in Wake County, collected via mail. Using the information from the housing sale data, the results of the survey can be matched to the hedonic data.
Four themes will be considered in this survey: (1) perceptions of the neighborhood and locational amenities relevant to housing choice decisions; (2) water-based recreation, including local and more distant recreation trips; (3) perceptions of water quality and the link between physical measures and other visual indicators; and (4) time allocation and the factors influencing the opportunity cost of time relevant to recreation models.
In addition to these attributes, we expect to collect a detailed set of socioeconomic characteristics of the respondents.
Two focus groups have been conducted. The first (July 23, 2002) considered the issues in eliciting information about recreation choices, the meaning of neighborhood and spatial amenities, and water quality perceptions. The second focus group (October 8, 2002) focused on local recreation outings, time allocation, and water quality perceptions.
After the first focus group was completed, a draft questionnaire was prepared. It has been subjected to three revisions to date: one before the second focus group, and two after the summary of findings from the second group were discussed. An analysis plan for the data to be collected was outlined. Based on this analysis plan, further revisions to the questionnaire were underway at the close of the project year.
A few key conclusions emerged from each focus group. They are summarized for each session individually:
Focus Group 1: (July 23, 2002)
· Factors in Households' Locational Choices. Job and employment are dominant considerations in household choices. Proximity to family, and weather and climate also were important factors. Highway congestion also was noted; recreation was a consideration but not a dominant factor.
· Household Concept of Neighborhood. Most participants identified neighborhood with housing subdivision, provided the area was not too large or subject to periodic expansion.
· Features of Household Recreation Choices. Time, rather than distance, was the key consideration in defining what participants meant by local recreation. Participants would not travel more than 2-3 hours for a day trip.
· Perceptions of Water Quality. Participants had wide differences in their perceptions of water quality. Neither exercise performed well in eliciting clear-cut responses. The responses of the first focus group motivated the questions used in the second focus group to obtain valuation information. Most participants could be classified into using either visual inspection or relying on authorities (i.e., the government) to provide information.
Focus Group 2: (October 8, 2002)
· Use of Time. Participants could describe time allocations; many thought that there were various local recreation activities that could be done in short periods of time. Some could use periods as short as 1 hour. The activities would vary with the amount of time available.
· Description of Recreation. Participants made a clear distinction between local outings, day trips, and longer. Participants would distinguish between them. Time was key constraint for local recreation. Planning increases with length of trip.
· Water Quality Perceptions. People used photos to describe water quality ratings affected by use; previous visual experiences affected ratings. A variation on Resources for the Future's water quality ladder seemed to provide an effective approach to eliciting water quality perceptions.
· Overall Judgment. The second group resolved key issues for surveys including: (1) a description of neighborhoods, (2) specification for factors in location choice, (3) a vehicle for water quality perception, and (4) a definition for local recreation.
Task 3: Development of a Water Quality Database
Water Quality Data have been assembled from a variety of sources, and all data are geo-coded. These measures are not available for all stations in all years. We are in the process of developing a single unified database to evaluate the coverage of the water quality data for the watersheds in Wake County.
At the close of the year, the team had agreed upon a format for the data sources, time spans, and general information assembled for the project, and the Ph.D. research assistant had outlined a strategy for merging the data in a GIS compatible format.
Task 4: Preliminary Analysis of Water Quality/Land Conversion Models as Part of Companion Research (Palmquist and Phaneuf)
Closely related research on the effects of land use conversion on water quality is nearing completion. The research is developing and estimating models to predict where development takes place, and what effect development has on nearby streams, lakes, and rivers. This research also focuses on Wake County, so that the results may be used in the current project.
The water quality model is using spatial econometrics to estimate the effects of past and current development within small geographic units (mini-HUCs) on objective measures of water quality in the adjacent waterways. We are controlling wastewater treatment plant discharges and rainfall, as well as water quality in the upstream waterway (spatial lags). We are using phosphorous, total nitrogen, total suspended solids, and the pollutants most likely to be influenced by development. The results seem to be quite robust and in accordance with our expectations.
We also are using a discrete duration model to explain the development decision using land and location characteristics as well as characteristics such as the presence of a sewer system. We intend to integrate the water quality and duration models in the near future.
Task 5: Conceptual Modeling for Locational and Joint Hedonic/Recreation Models
Currently, Dr. Smith is working with a graduate student (Jaren Pope) supported by the project. They are investigating the neighborhood definitions using simple hedonic property value models. The objective is to evaluate the range of price indices developed for alternative neighborhood definitions, and the consistency of their ranking with the ranking of local attributes conveyed through location in neighborhoods (i.e., following the logic of Drs. Sieg, Smith, Banzhaf, and Walsh, Journal of Urban Economics, July 2002).
Drs. Palmquist, Phaneuf, Pope, and Smith have initiated the process of developing extensions to the McConnell (Land Economics 1990) and Parsons (Land Economics 1991) hypothesis that hedonic and travel cost models should be jointly determined. Our focus has been on the types of recreation that likely are to be directly affected by residential location.
Future activities for this research project include the following:
· Survey of Households. The final mail survey will be implemented in early 2003. We expect work to require 2 to 3 months of effort from printing to multiple mailings and data entry.
· Recreation Data. We contacted Dr. V. Robert Leeworthy to acquire the 2000 National Recreation Survey. He has agreed to provide the data requested. We expect to finalize the acquisition of these data and start work with them on the random utility model (RUM) component of the analysis.
· Locational Equilibrium Modeling. With the assistance of Dr. Daniel Hallstrom (an Agricultural Resource Economics faculty member), a preliminary, simpler program for estimating preference parameters for the locational equilibrium model has been developed. We expect to refine this code and to initiate comparisons of alternative models.
· Local Recreation and Residential Choices. We expect to extend our initial conceptual modeling in this area and use the survey data, along with the housing data, to develop tests of the most direct hypotheses implied by the model.
· Hedonic Modeling. Dr. Palmquist has added a detailed array of location-specific data to the Wake database. Soil characteristics and watershed measures had been linked via GIS. We expect to enhance initial hedonic models with these new data.
Supplemental Keywords: revealed preference, non-market valuation, random utility, hedonic property value, household choice, water quality value, land conversion, behavior model, benefits assessment, economic benefits, ecosystem valuation, efficient household framework, environmental values, household choice, land cover changes, measuring benefits, model aggregation methods, residential property values.