NSW Department of Environment and Climate Change Envalue Study Database

Improved Air Quality
Benefit transfer of primary estimates may be undertaken if the criteria for benefit transfer are satisfied. The similarity of the environmental characteristics and the proposed environmental change at the study and policy sites are of a particular importance when transferring direct estimates of the benefits of improved air quality. Although study and policy sites might use similar terms for subjective indicators of air quality (such as poor/fair/good), they may not be comparable. Also, people in different parts of the world, from different cultures, and accustomed to different levels of air quality, are unlikely to have the same definition for either the base air quality condition (say, poor air quality) or the contemplated improvement. In view of such problems, transfer of primary estimates may involve considerable uncertainties.

Valuing the Health Impacts of Air Pollution
The suggested methodology for transferring estimates from study to policy sites is to transfer dose-response relationships. If the less robust alternative of transferring aggregate benefit estimates is adopted, then substantial adjustments will need to be made to take account of differences in the ambient concentrations of pollutants, population and other demographic and socioeconomic characteristics of the study and policy sites. The suggested methodology for transferring dose-response relationships is as follows: 1. Determine the change in air pollution at the policy site 2. Determine the likely health effects and the number of people exposed to excessive levels of air pollution 3. Assess the effect of any qualifying factors 4. Apply dose-response relationships 5. Compute the benefit of improving air quality given a range of values for morbidity. The factors that need to be considered when transferring dose-response relationships and marginal per-person benefit estimates include: Meteorological differences: the impact on dose-response relationships of differences in temperature and humidity Antagonistic/synergistic factors: differences in the concentrations of other pollutants at policy and study sites and their impact on the health effects of the pollutant in question Confounding factors: for example, differences in work-related exposure to air pollutants and smoking Long-term impacts: there is relatively little information reported on long-term impacts. A number of pollutants lower life expectancy rather than (or in addition to) causing acute symptoms or immediate death Thresholds: there is dispute about whether safe thresholds for the concentration of certain pollutants exist. Current evidence shows that the thresholds, if they exist at all, are much lower than previously thought Non-representative demographic sample: many of the studies have been of adult, urban, working males. Often a disproportionate part of the sample includes people with impaired respiratory functions to demonstrate more vividly the effects of air pollution. In these cases it is not possible to extrapolate from the sample response to the population as a whole Correlation with unmeasured effects: Many primary studies only examine the impacts of a few pollutants. In studies that examine a small number of pollutants it is possible that the estimated impacts would include, in part, the impacts of other pollutants.

Agricultural Impacts of Air Pollution
As with health benefits from air quality improvements, when transferring estimates from study site to policy site the preferred approach is to use dose-response or `yield-decrement' relationships. The relationship between air pollution and yield is complex. A number of factors determine the impact of air pollution on the value of agricultural yield, eg. crop type and variety, soil type, wind, rainfall, water stress, temperature, management practices and the market value of produce. The impact of air quality on agricultural yields is largely site-specific because of the complexity of these relationships. The methodology suggested for transferring yield-decrement relationships is as follows: 1. Determine the change in air pollution 2. Determine the types and varieties of crops grown and the amount of each 3. Assess the effect of any qualifying factors 4. Apply yield-decrement relationships 5. Compute the change in dollars using current market prices Factors which may systematically affect the yield response of crops to air pollution, and which are therefore relevant for benefit transfer, include: Environmental Factors - In most controlled trials, environmental conditions such as the amount of water applied and temperature are held at levels which maximise yield so yield-decrement relationships can be generated. However, conditions at many policy sites may be less conducive to crop growth than those in controlled trials. In general, when plants are under stress from other factors (such as high or low temperature, wind, or low moisture), they are more susceptible to damage from air pollution. The benefits to agriculture of improved air quality will tend to be underestimated in these cases. With some plants, however, moisture stress may actually reduce the impact of air pollution. This is because the plants respond to inadequate moisture by partly closing their pores to reduce water loss. A side-effect of this action is that the plants absorb and release lower amounts of carbon dioxide and other gases. This mechanism was estimated by Adams, Hamilton & McCarl (1986) to reduce the effects of ozone on corn, soybeans, wheat, cotton, grain sorghum and barley by about 25%. Peak Concentrations - The relative importance of peak vis-à-vis mean ambient concentrations differs with each pollutant. For ozone it has been found that peak episodes of pollution cause corresponding bands of necrotic or dead tissue to appear, suggesting that peaks are more important in affecting yield (Leece, 1993). For fluoride it appears that mean ambient concentrations are more important, as plants build up fluoride concentrations throughout the growing season and symptoms are related to leaf concentration. In the case of sulfur dioxide, studies have shown that long-term mean rather than peak concentrations are more important for determining the effects on yield. However, intermittent acute exposures do produce different plant responses (Murray & Wilson, 1989). The dose-response relationships presented in the Attachment are based on mean ambient concentrations, reflecting the general availability of data. Heck et al. (1984) estimate the effect of both mean and peak ambient concentrations of ozone on yield, but contend that the peak estimates are not transferable, without explicitly explaining why not. The use of dose-response relationships based on mean concentrations is less than ideal when peak levels of a pollutant are the more important consideration. One way to improve the existing dose-response relationships would be to weight more heavily the periods in which plants are most sensitive to air pollution, such as during germination and on days when temperatures are high. Antagonistic/Synergistic Effects - Often the impact of one pollutant on yield will depend on the presence of other pollutants. For example, Amundson (1983) found significant differences in soybean yield at high concentrations of sulfur dioxide when the plants were also exposed to nitrogen dioxide. Irving (1983) reports that the impact of acid rain on the yield of alfalfa, fescue and mustard green was found to differ with different sulphate:nitrate ratios even though the pH remained constant. Hence it is important to consider the levels of background pollutants and whether they will have antagonistic or synergistic effects. Substitution Effects - Substitution effects are generic to all impacts on agriculture. They arise because producers can usually alter their production process to reduce the impact of high ambient concentrations of pollutants on crop yield. For example, producers can adjust their crop mix by planting crops that are less sensitive to air pollution. The effectiveness of such an adjustment will depend partly on whether the new crop mix produces a similar gross margin. In terms of benefit estimation, substitution effects will generally reduce the benefit of improved air quality. Higher Costs - The effects of changes in enterprise and yield on both variable and capital costs need to be considered. Changes in yield and crop mix will affect variable as well as capital costs. Capital costs may change because different crops have been planted which, in certain cases, may require the use of different or additional machinery. Elasticity of Product Demand - Agricultural markets are often characterised by elastic product demand, in which an increase in total yield will lead to a fall in price. This is commonly ignored and analysis is restricted to partial equilibrium. A number of commentators have recently argued that accurate estimation of benefit requires analysis of price effects (eg. Kopp et al., 1985; Adams, 1986; Kopp & Krupnick, 1987). If the increased yield resulting from improved air quality is large enough to affect the market price; it will reduce the benefits to producers of improved air quality. Conversely if air quality declines, producers could be made better off if the price increase were to offset a reduction in output. However, consumers are always made worse off when prices increase, and overall there will be a net decrease in welfare.

Water Quality - Drinking Water
As the definitions of water quality categories may differ between sites, and variation within categories is also possible, it is advisable to give careful attention to the similarity of site characteristics when using benefit transfer. The existence of substitute water sources (eg. rainwater tanks) may also substantially influence the respondent's WTP.

Recreational Water
As the definitions of water quality categories may differ between sites, and variation within categories is also possible, it is advisable to give careful attention to the similarity of site characteristics when using benefit transfer. The existence of substitute water sources (eg. rainwater tanks) may also substantially influence the respondent's WTP. Different categories of water quality for recreational use have been defined by a number of commentators. Swimmable Fishable Boatable Non-Boatable While this categorisation is convenient, the way people judge whether water falls within a particular category remains subjective, and this can be expected to lead to variation between studies. As is the case with drinking water, benefit transfer is possible where the criteria for benefit transfer are satisfied. When transferring estimates, factors to be considered include: Whether a river is urban or rural: rivers in an urban setting are likely to be valued differently from rivers in the country. Substitute sites: especially if used for recreation, the existence of substitute, less-polluted recreation areas will reduce the values placed on water quality improvements. Importance to the community: if the river or lake is important to the economy of the area because of its value for recreation, then water quality might be valued more highly.

Water Quality - Salinity - Domestic Impacts
Households incur costs form salinity of the domestic water supply in many areas of Australia. The service lives of domestic capital goods such as water pipes and fittings, hot water systems and domestic appliances are reduced due to corrosion and scaling in pipes. Increased ongoing costs result from increased maintenance, and the need to use water softeners and extra soap and detergent. In the case of households, damage from salinity arises from increased total hardness (TH) and corrosiveness of the water supply. Errors may be introduced when transferring estimates because of difference in water quality parameters. In particular, while water may have the same level of salinity (as measured by electric conductivity or by concentration of total dissolved solids), TH and the concentrate of chloride ions may be different. Variation may also result where there are difference between study and policy sites in material used for water pipes, taps and water heaters. Such difference would be expected tot be more pronounced for primary studies that are more than 10 years old. Water Quality - Salinity - Agricultural Impacts. The suggested methodology for transferring yield-decrement relationships to estimate the costs of the impact of salinity on agricultural yield is as follows: 1. Determine the salinity of the irrigation water 2. Determine the types and varieties of crops grown and the amount of each 3. Determine the soil types 4. Assess the effect of qualifying factors 5. Transfer yield-decrement relationships 6. Apply current market prices. Factors which may systematically affect the yield response of crops to salinity, and which therefore are relevant for benefit transfer, include the factors listed below. Environmental factors - Environmental conditions are held constant at levels that optimise plant growth in most of the trials on which the yield-decrement relationships are based, yet at many policy sites the conditions may not be optimal for crop growth. In general, when plants are undergoing stress from other factors (such as high temperature or waterlogging), they are more susceptible to damage from salinity. The effect of spikes - For many crops, the effect of salinity on yield will depend on the significance of salinity `spikes (short periods of particularly high salinity), as opposed to average salinity levels. Spikes are often caused by localised downpours which fall in upstream areas and wash salt deposited on the soil surface into streams, but which are not large enough to substantially increase the flow of the main stream. They may also be caused by the release of large amounts of saline mine water. The yield-decrement relationships presented in the database are based on average salinity levels. There is insufficient information presently available to determine yield-decrement relationships due to peak salinity levels. Variation in salinity during the growing season - Crops are most sensitive to spikes during germination and early seedling stages. This suggests it would be useful to create a model that consists of seasonally weighted yield-decrement relationships to take account of the period when crops are most sensitive to stress from salinity. Weighting could be attached to simulate the impact of salinity on final yield during each of these periods by dividing the growing season for each crop into a number of periods. Such a model would produce better estimates than the traditional `bent-stick approach for crops such as grapes, stone fruit and vegetables, which are particularly sensitive to salinity during specific periods. The bent-stick model derives its name because there are thresholds for each crop after which yield declines linearly with increasing salinity. When graphed the relationship looks like a bent stick. The development of such a model is the natural next stage in the refinement of salinity models. It would be extremely useful for evaluating the benefits of schemes where control can be exercised over salinity levels for restricted lengths of time. For many crops it is suspected that the information to support such a model already exists. An intermediate refinement is to use the existing yield-decrement relationships, but to give extra weighting to salinity during critical periods. Dwyer Leslie (1992) used such an approach in the Kerang Lakes Agricultural Economic Impact Model. Irrigation water salinity was averaged for stone fruit from 1 August to 14 November, Oranges from 1 October to 31 December, and grapevines from 14 October to 14 December. Salinity levels during the critical growing period were given twice the weight of average salinity during the year to derive the effective average yearly salinity level. Time period - The duration of the increase in salinity is important for determining the impact of salinity on yield. Soil water takes several years to reach an equilibrium level of salinity. Also grapevines and some other plants appear to be able to recover quickly from temporary periods (up to several years) of salinity. So when the change in salinity levels lasts only for a few years, there will be a smaller impact than if the changes are permanent. Higher or lower costs - The effects of changes in crop mix and yield on both variable and capital costs need to be considered for their effect on farm profits. Changes in crop mix and yield can affect variable as well as fixed costs. For example, they may require changes in fertiliser use, water requirements and machinery. Elasticity of product demand - Agricultural markets are often characterised by `elastic product demand': that is, an overall increase in yield will often lead to a fall in price. Where this occurs, it is likely to reduce the benefit of a reduction in salinity. However, in general, only partial equilibrium analyses have been undertaken and price effects have not been considered. Irrigation technique - Both the method and the timing of irrigation can cause significant variation in yield-decrement relationships. For example, stone fruit are more sensitive to overhead irrigation and crops are generally less sensitive to salinity if they are irrigated at night. Substitution effects - High salinity levels can preclude the planting of certain crops that would otherwise have a higher net value. It follows that reduction in salinity levels may allow the substitution of higher value but more salt-sensitive crops that provide higher returns at lower salinity levels. The benefits from reductions in salinity will be decreased when this substitution is taken into account. Water Quality Salinity - Industrial Impacts Many industries require water with low salt and mineral content for their production processes. Low-salinity water is required for steam generation, water cooling and for use as process water, though in the case of heat exchangers, the design and operation can be modified (at a cost) to use water of higher salinity. Estimates of the industrial cost of increased salinity are site-specific. They depend on both the number and type of industries that use water as an input, and their efficiency. Therefore, the transfer of estimates form study site to policy site will need to take explicitly account of difference between these sites. The estimates reported in the database aim only to indicate the order of magnitude of the estimates made and to illustrate the methodology used. Dryland Salinity Several estimates have been made of the costs of dryland salinity in Australia. The Standing Committee on Soil Conservation (1982) estimated that productivity losses from dryland salinity amounted to $26 million per year; while Peck, Thomas & Williamson (1983) estimated that the total annual damage cost to agriculture was $35 million. Estimates that may be used for benefit transfer are available for the impacts listed below. Lost production - Observed differences in yield are often used to estimate the costs in lost production from dryland salinity. Typically, dryland salinity due to rising watertables occurs over small areas in what are known as `discharge' areas. These areas often adjoin agricultural areas that are unaffected by salinity. The costs of dryland salinity can be estimated by comparing the difference in farm returns per hectare between affected and unaffected areas, and relating this to the severity of dryland salinity. Decline in property values. The decline in the market value of land (hedonic price method) may provide information on the costs of dryland salinity. Water quality. The costs of salinisation of domestic, agricultural and industrial water supplies are well documented. Dose-response relationships for the effect of salinity on domestic and agricultural water use are examined in Section 5.5. Ecosystems. Dryland salinity affects aquatic ecosystems and wetlands. The preservation value of wetlands and lakes is dealt with by some of the studies in ENVALUE. Aesthetic and recreational damages. These relate to losses in enjoyment from using sites that have been damaged by salinity. Estimates of the recreational value of wetlands, lakes and rivers are reported in Section 8, Natural Areas.

Land Contamination
The following factors should be considered when assessing whether the criteria for benefit transfer are satisfied: whether the site is in an urban or rural area and, if in an urban area, whether it is close to an urban centre whether members of the community perceive a health risk socioeconomic characteristics (eg. households with children generally have a greater fear of health risks) proximity of high-value housing whether the market at the study site has capitalised the impacts from the hazardous waste site similarity of operating and long-term impacts between the study site and the policy site (eg. odour from landfill gas, groundwater contamination) A limitation of using the hedonic price method to estimate the economic cost of complex environmental hazards is that it only produces consistent results when people are aware of the changes in environmental quality. Several commentators have found that the impact of a hazardous waste site on housing prices varies over time due to differences in perceptual cues that convey information about health risk. These perceptual cues include odour and pooled leachate, as well as non-environmental cues such as increased media interest about health risks. McClelland et al. (1990) report that, following the closure of the Operating Industries Inc. landfill in Los Angeles and the subsequent cessation of odours and elimination of leachate pools, people thought there had been a large reduction in the health risk. Kohlhase (1991) found that house prices fell only when the sites considered were announced by the US EPA as being on the priority list for remedial action, thus making people aware that there actually was a health risk.

Aircraft Noise
The following factors may affect estimates when undertaking benefit transfer with reference to aircraft noise: whether the policy or study sites contain predominantly high or low price properties whether background noise will mask additional noise and whether the existence of background noise levels at either the study or policy site have affected housing prices to the extent that they are likely to be less responsive to additional noise impacts whether quiet housing is not available or the housing market is distinctly segmented into low-price and high-price areas. If either or both are the case, it is unlikely that differentials in housing prices will accurately indicate the impact of noise whether housing prices at the study or policy sites are subject to high inflation. If so, this can distort benefit estimates.

Road Transport Noise
The following factors should be considered when transferring benefit estimates in addition to the factors mentioned under aircraft noise: Noise may affect higher priced houses more significantly (eg. Palmquist, 1980; Holsman & Bradley, 1982). Unit increases in noise may cause higher reductions in house prices at higher levels of noise (eg. Nelson, 1980; Holsman & Bradley, 1982). Therefore care should be taken when mean impacts are extrapolated to significantly higher or lower noise levels. Differentials in housing prices rarely capture only the effect of noise because of multicollinearity. A number of studies have attempted to isolate the effects of air pollution; however variables such as air pollution have generally been omitted because of problems with collinearity. In practical terms, the reported estimates are likely to include effects from other factors in addition to noise. The existence of air pollution, danger from traffic and amenity losses from living near major roads is likely to cause the impact of noise to be overestimated if there is collinearity between included and omitted variables.

Radiation
The following factors should be considered when transferring benefit estimates: the average and range of exposure levels in study and policy sites differences in per capita incomes and levels of health care costs.

Natural Areas
Natural areas show great diversity in their characteristics and in the activities they support. Many natural areas have unique features. It may not always be possible to select study sites that are sufficiently similar to the policy site. And differences in values placed on comparable natural areas may result from cultural differences. This effect is likely to be most pronounced when estimates are being transferred between countries, but may be relevant even within Australia. For example, many Sydney residents attach importance to beaches because they are an important part of their lifestyle. In view of these considerations, the transfer of values from individual study sites can be difficult and subject to considerable uncertainty. The preferred approach therefore is to transfer complete demand functions, or a range of estimates from similar sites. Depending on what is measured, most valuations are expressed in terms of per household per year or per visitor day. Factors relevant to transferring estimates of the value of natural areas include: in the case of preservation values, defining the alternative to preservation. In some studies a particular impact may be specified, eg. the construction of a geothermal power plant. Such a development is intrusive in an area of natural beauty but may still allow some activities to continue, though with lower enjoyment. In other studies, the alternative to preservation is the virtual loss of the site as a natural area in survey methods the type of sample selected can affect the estimates obtained. For example, differences may be expected in the responses between residents and visitors. Additional factors that should be taken into account in surveys include respondents' level of knowledge of the area and of the proposed environmental impact, and the distance they live from the site the availability of substitute sites. Rather than transferring valuations from individual primary studies, a more comprehensive approach has recently received some attention. This involves reviewing reported estimates in a statistically consistent way by undertaking an econometric analysis of the relative importance of various factors in estimates of environmental values. This approach has been referred to as `meta-analysis'. Smith & Kaoru undertook such an investigation to assess the extent to which behavioural assumptions and model simplifications made by the analyst had influenced the estimated values, compared with the magnitude of genuine effects. Their analysis indicated the importance in the travel cost method of the opportunity cost of travel time and of including substitute sites when specifying models. While they consider that meta-analysis shows considerable promise, they conclude that, at this stage, `it would not be prudent to recommend this type of model for predicting [consumer surplus] for policy analysis. Walsh et al. used regression to estimate the influence of different variables in determining recreational demand for particular sites. Their models explained 39% of the variation in consumer surplus estimates for the travel cost method and 44% for the contingent valuation method. They found that omission of the costs of travel time reduced the estimated travel cost method benefits by around 30%; and that, on average, estimates that omit cross-price terms for substitute sites could be lower by up to 30%. Average values per recreational day varied significantly for different activities, ranging from $17 for picnicking to $72 for salt water fishing, around an overall mean value of $34.

Bibliography (Benefit Transfer Text)
Adams, R.M., Hamilton, S.A. & McCarl, B.A. (1986). `The Benefits of Pollution Control: the Case of Ozone and US agriculture'. American Journal of Agricultural Economics, 68(4): 886-893. Amundson, R.G. (1983). `Yield Reduction of Soybean Due to Exposure to Sulfur Dioxide and Nitrogen Dioxide in Combination'. Journal of Environmental Quality, 12(4): 454-459. Desvousges, W.H., Naughton, M.C. & Parsons, G.R. (1992). `Benefit Transfer: Conceptual Problems in Estimating Water Quality Benefits Using Existing Studies'. Water Resources Research 28(3): 675-683. Dwyer Leslie (1992). Kerang Lakes Area Management Plan Agricultural Economic Impact Model. Prepared for Rural Water Corporation, Victoria. Heck, W.W., Cure, W.W., Rawlings, J.O., Zaragoza, L.J., Heagle, A.S., Heggestad, H.E., Kohut, R.J., Kress, L.W. & Temple, P.J. (1984). `Assessing Impacts of Ozone on Agricultural Crops: II. Crop Yield Functions and Alternative Exposure Statistics'. Journal of the Air Pollution Control Association, 34: 810-817. Holsman, A.J. & Bradley, R. (1982). The Economic and Social Impact of Main Road Traffic Noise in Sydney. Report Prepared for the NSW State Pollution Control Commission, Sydney. Irving, P.M. (1983). `Acidic Precipitation Effects on Crops: A Review and Analysis of Research'. Journal of Environmental Quality, 12(4): 442-453. Kirchoff, S., Colby, G., LaFrance, J. T., (1997), Evaluating the Performance of Benefit Transfer: An Empirical Inquiry, Journal of Environmental Economics and Management, 33, (75-93) Kohlhase, J.E. (1991). `The Impact of Toxic Waste Sites on Housing Values'. Journal of Urban Economics, 30: 1-26. Kopp, R.J. & Krupnick, A.J (1987). `Agricultural Policy and the Benefits of Ozone Control'. American Journal of Agricultural Economics, 69(5): 956-962. Kopp, R.J., Vaughan, W.J., Hazilla, M. & Carson, R. (1985). `Implications of Environmental Policy for US Agriculture: The Case of Ambient Ozone Standards'. Journal of Environmental Management, 20: 321-331. Leece, D. (1993). Personal Communication. NSW Environment Protection Authority, Bankstown. McClelland, G.H., Schulze, W.D. & Hurd, B. (1990). `The Effect of Risk Belief on Property Values: A Case Study of a Hazardous Waste Site'. Risk Analysis, 10(4): 485-97. Murray, F. & Wilson, S. (1989). `The Relationship Between Sulfur Dioxide Concentration and Yield of Five Crops in Australia'. Clean Air, 23(2): 51-55. Nelson, J.P.(1980). `Airport Noise and Property Values: A Survey of Recent Evidence'. Journal of Transport Economics and Policy, 14: 37-52 OECD (1992) Project and Policy Appraisal: Integrating Economics and Environment (authors Pearce, D., Whittington, D., Georgiou, S. and James, D.) OECD (1994) Transferring Benefit Estimates, Chapter 10 in Project and Policy Appraisal: Integrating Economics and Environment. OECD. (Authors Pearce, D., Whittington, D., Georgiou, S and James, D.) Palmquist, R.B. (1980). Impact of Highway Improvements on Property Values in Washington. National Technical Information Service, Springfield, Virginia, in Nelson, J.P. (1982). `Highway Noise and Property Values, A Survey of Recent Evidence'. Journal of Transport Economics and Policy, 117-138. Peck, A.J., Thomas, J.F. & Williamson, D.R. (1983). Salinity Issues Effects of Man on Salinity in Australia. Water 2000: Consultants Report No.8. AGPS, Canberra. Standing Committee on Soil Conservation (1982). Salting of Non-irrigated Land in Australia. Department of Primary Industries, Canberra, in Yapp, T. (1989). `The Costs of Degradation to the Community: Issues and Estimates'. Australian Journal of Soil and Water Conservation, 2(3): 32-36.