Two benefits transfer applications (Pennsylvania, Georgia, Maine)

Environmental policy is often conducted under both time and information resource constraints. This combination necessitates the use of benefits transfers where policy makers take estimates of economic benefits from prior studies and applications and use these value estimates as proxies in a different context. This thesis examines two different aspects of benefits transfers: an analytical approach and a methodological focus. The raw data for these exercises comes from two separate surveys examining household demand for clean water as measured by willingness to pay. The first, conducted in 1996, estimated household willingness to pay for protection from nitrate contamination in Georgia, Maine and Pennsylvania using the contingent valuation approach. The second collected data about household preferences for cleaning up rivers and streams from acid mine drainage in three different areas within Pennsylvania using the attribute based stated choice model.; The two main objectives are to compare the different methods of benefits estimation and transfer—contingent valuation and attribute based stated choice models; and to compare two modes of analysis—multivariate regression and artificial neural networks.; Methodologically, the attribute based stated choice model has begun to gain favor with researchers due to its flexibility. These research results confirm the flexibility as well as identifying a scaling property or confidence rating variable that is extremely useful in terms of identifying the strength of an individual's choice. Difficulties with the stated choice approach are also discussed.; The analytical focus compares the predictive performance of artificial neural networks (ANN) to multivariate regression. The results show that, in terms of predicting choice responses, ANN are promising. Unfortunately, the out of sample forecasting properties are extremely data dependent. However, results from ANN may be used as a supplement to multivariate techniques for identifying difficult to detect nonlinearities in data.; The most important contribution of this work is the identification of two ways to measure how close benefits transfers need to be in order to be policy effective. The first measure shows that in the cases examined here it is likely that the analyst may make the wrong decision in implementing a program. The second measure however, makes the job of the analyst easier by identifying certain cost thresholds for using transfers.