You are here:
The Role of Locational Equilibria and Collective Behavior in Measuring the Benefits of Air Pollution Policies
Objectives/Hypotheses: This research proposes to use the spatial equilibrium in estimating the economic value of reductions in air pollution. This equilibrium includes recognition of households' choices of residential locations based on site characteristics, including environmental amenities, as well as the collective choice process determining the amount of community provided local public goods. This spatial equilibrium estimator (SEE) allows estimates of benefits from air pollution policy to take account of general equilibrium adjustments in response to policy, and permits a behaviorally consistent description of a policy's distributional consequences (including environmental justice and racism). The research has three objectives: (1) to extend the SEE framework to include environmental public goods and to allow for a more detailed characterization of the observable sources of heterogeneity in household preferences; (2) to apply the SEE framework along with conventional hedonic property value and simple (i.e., mulitnominal logit) RUM frameworks using current data for air quality (primarily ozone, nitrogen oxide, and particulate matter), housing prices and characteristics as well as community characteristics, in the Los Angeles area; (3) to compare the benefit estimates implied by each framework for a set of current policy alternatives and their respective sensitivities to their maintained hypotheses. Approach: To meet the objectives we will construct a data set for a set of counties in California that combines complete information on residential property values with the Air Resources Board and EPA ambient air quality data as well as information about local schools and neighborhood characteristics from census statistics. The geographic resolution of the housing information includes longitude and latitude as well as other identifiers to permit the merge. Maximum likelihood and simulation estimators will be applied to the sample at different levels of spatial and temporal resolution. The comparison of results will focus on the estimates of marginal willingness to pay. Expected Results: The research should: (a) evaluate the importance of local public goods for the measurement of the benefits of reducing air pollution; (b) compare the SEE with the simpler revealed preference hedonic and RUM approaches; and (c) provide new estimates of the incremental benefits for reducing air pollution in Southern California that can serve as plausibility checks for EPA's effects - specific measures. Improvement in Risk Assessment or Risk Management: EPA's current benefit practices monetize the economic values derived from each specific risk reduction. As a result, each effect is treated as independent (for valuation), and their sum can be a larger amount than an individual would be willing to pay for the composite of the risk changes as a group. Benefit measures based on locational adjustments provide a plausibility check for the effect-specific measures.
R826609Principal Investigators: Smith, V. Kerry
Sieg, HolgerTechnical Liaison:Research Organization:
Duke UniversityFunding Agency/Program: EPA/ORD/ValuationGrant Year: 1998Project Period: September 1, 1998 - August 31, 2001Cost to Funding Agency: $199,948
- Project Reports
O'Sullivan A, Sexton TA, Sheffrin SM. Property Taxes and Tax Revolts: The Legacy of Proposition 13. Cambridge, UK: Cambridge University Press, 1995.
Epple D, Sieg H. Estimating equilibrium models of local jurisdiction. Journal of Political Economics 1999a;107(4):645-681.
Willig RD. Incremental consumer's surplus and hedonic price adjustment. Journal of Economic Theory 1978;(17)2:227-253.
Sieg H, Smith VK, Banzhaf HS, Walsh R. The role of optimizing behavior in willingness to pay estimates for air quality. American Journal of Agricultural Economics (in press).
Brunner E, Sonstelie J. Coping with Serrano: voluntary contribution to California?s local public schools. Unpublished paper, Department of Economics, University of California, Santa Barbara, CA, October 1996.
Sieg H, Smith VK, Banzhaf HS, Walsh R. School and air quality and their impact on residential choices in equilibrium: evidence from Southern California (working draft, August 1999). To be presented to the Econometric Society.
- Project Status Reports
For the year 1999
This research proposes to use the characteristics of a spatial equilibrium in estimating the economic value for reductions in air pollution. This equilibrium includes recognition of households' choices of residential locations based on site characteristics, including environmental amenities, as well as the collective choice process determining the amount of community-provided local public goods. This spatial equilibrium estimator (SEE) allows estimates of benefits from air pollution policy to take account of general equilibrium adjustments in response to policy, and permits a behaviorally consistent description of a policy's consequences. The research has three objectives: (1) to extend the SEE framework to include environmental public goods and to allow for a more detailed characterization of the observable sources of heterogeneity in household preferences; (2) to apply the SEE framework along with conventional hedonic property value and simple (i.e., multinomial logit) RUM frameworks using current data for air quality primarily ozone, nitrogen oxide, and particulate matter), housing prices and characteristics as well as community characteristics, in the Los Angeles area; and (3) to compare the benefit estimates implied by each framework for a set of current policy alternatives and their respective sensitivities to their maintained hypotheses.
Progress Summary: Task 1: Develop and Evaluate Extension to Spatial Equilibrium Estimator. The research completed for this task during the first year has involved three sets of activities:
- Review of the recent literature on the modeling of local public education and the role of peer group effects in the context of Tiebout and locational choice models, and on the effects of Proposition 13 and the reform of local funding of education in California and its implications for our proposed model
- Derivation of the welfare implications of the SEE's boundary indifference and stratification conditions, and linking the single crossing property with the conditions assumed in conventional nonmarket evaluation.
- Extension of the SEE framework to allow for three random sorting variables?income, student ability, and unobserved heterogeneity in the taste for local public goods.
Task 2: Assemble Data for Modeling. The majority of the data development has been completed, including:
- A complete record of the residential home sales from 1987 to 1995 for Ventura, Los Angeles, Orange, Riverside, and San Bernardino Counties in California.
- Complete air pollution readings for this area by monitor over the same period.
- Detailed records on per public educational expenditures and test scores by school districts for 1992-1993.
- Sales and housing characteristics information in a format that allowed development of a statistical database, including latitude and longitude for each house and monitoring station.
- Development of a computer code to match closest distance between the home and air pollution monitor readings to housing sales.
By the close of the first year of the project, complete data sets for Ventura, Orange, Riverside, and San Bernardino Counties were completed. Preliminary hedonic and SEE estimates had been developed for a subset of the areas (defined by a geographic boundary?the approximate natural topography of the air shed). Los Angeles, the largest county in Southern California, also was virtually complete. It was expected by mid-October that housing, air quality, local school test scores, income distributions, and census variables would be matched to the relevant data sets for estimation. Analysis will focus on the early 1990s, where we have overlapping information.
Task 3: Estimation of SEE Framework With Southern California Data. We have completed a preliminary analysis of the SEE model with information developed for 36 school districts from Orange, Riverside, and San Bernardino Counties. Completing this research required us to "test" the matching framework for the housing and air pollution data as well as to merge the test scores (for the education measure) and the income distribution records.
It also required estimation of an area wide hedonic model. For comparison, we developed a separate hedonic model with air pollution and other location-specific variables included. The SEE framework was extended to include peer group effects. A preliminary version of the model has been estimated using the 36 school district sample.
At the close of the first year of the project, we had nearly completed the analysis of a preliminary housing price index, estimation of the income distributions, and school quality measures for the 92 school districts to comprise our full sample (including Ventura, Los Angeles, Orange, Riverside, and San Bernardino Counties). We were in the process of expanding the air pollution measures to be used and the matching criteria for linking them to individual houses.
Task 4: Estimation of Hedonic and RUM Frameworks. Initial estimates of the marginal willingness to pay for reducing ozone developed using the SEE framework were compared to a simple hedonic model estimated for the initial 36 school district sample. The development of this comparison raised methodological and empirical issues; a hedonic model specified in terms of structural characteristics and lot size was used to develop school district-specific housing price indexes. Detailed hedonic models for nonmarket valuation, along with random utility models, cannot be estimated until the full data set is completed. This will be an important component of the second year's activities.
Task 5: Benefit Measurement and Evaluation. This research was planned for the third year of the project. Researchers initiated background research required to complete this task.
Conceptual Development. Smith has begun the task of identifying the conceptual features of the SEE estimator. Some of these are discussed in the first completed paper. He presented further results at Iowa State University. This research will continue? considering the implications of preference specification for the implied associations between air pollution and school quality measures. These findings are expected to be assembled in a paper on welfare measurement within the SEE framework.
Data Development and Estimation. Integration of the Los Angeles County information into the overall database is expected to be completed within the next 2 months. Re-estimation with SEE of our initial model also is planned with the expanded set of school districts. In addition, the issues associated with three aspects of the initial modeling effort?measures of an pollution and their links to housing, structural specification of the relationship between air quality measures and other public goods, and the index number problems posed in developing the housing price index?will be investigated.
Hedonic and RUM Estimation. How SEE, hedonic, and RUM indexes compare in developing local price indexes will be evaluated. This research should yield estimates of the other two methods. Separate comparisons of hedonic models over time and market definition will be developed. Several sets of results from these efforts are expected.
Table 1. Crosstabulation of Housing Transactions by Year and County
YearOrange CoRiverside CoSan BernardinoVentura Co.Total198711,4625,3417,1203,31927,242198811,1157,4471,4993,66223,72319897,55710,6019,4413,06730,66619909,6499,9989,7102,45231,809199112,2878,6909,5953,05333,625199212,5789,36910,1503,58135,678199314,91311,24411,4105,48643,05319949,38613,6483,5046,52653,064199516,45014,23214,2656,45951,406Total115,39790,57086,69437,605
YearLos AngelesOrange CoRiverside CoSan BernardinoVentura Co.Total19874236419198842554201989625652419907256525199162555231992735842719937271263419947261163219957265626Total5519476445230
Table 2b. Crosstabulation of Ozone Air Quality Monitors by Year and CountyYearLos AngelesOrange CoRiverside CoSan BernardinoVentura Co.Total1987185812952198816581385019891558108461990155611845199117591174919921759127501993155101375019941548137471995154812746Total143437411770447
Table 3a. Summary of Pollution Readings Across Monitor-Years (1987-1996)
Annual Average PM10 (g/m3)CountyNo. Obs.MinimumMeanMaximumStnd. Dev.Los Angeles6125.545.968.110.3Orange2130.139.750.35.7Riverside5326.850.894.518.4San Bernardino7519.846.281.318.2Ventura4722.130.641.04.6
Table 3b. Number of Exceedence of National 1-Hour Ozone Standard (ozone ppm)CountyNo. Obs.MinimumMeanMaximumStnd. Dev.Los Angeles155037.114834.7Orange47010.94110.9Riverside86033.312330.4San Bernardino117047.513042.9Ventura7005.2529.1