Theoretically optimal environmental metrics and their surrogates

Many environmental pollutants are thought to produce several classes of effects, e.g., morbidity, mortality, property damage. It would be possible to specify a unique theoretically optimal metric for each class of effects. Alternatively a single surrogate metric could be used to represent the combined impact of all classes of effects. Here, the use of surrogates is discussed in the context of regulation of ambient air pollution. It is demonstrated that costs may be associated with the use of surrogates, and that these costs are caused by (1) imprecision in the relationships between the surrogate and each of the theoretically optimal metrics, and (2) reduction in dimensionality of the set of exposure metrics. The factors which govern the magnitude of costs incurred are explored. It is noted that any savings inherent in the operation of simplified systems of monitoring and data management offset to some degree the costs generated by the use of surrogates.