input-output framework, but attempt to be more descriptive. These approaches take account of the environment by assembling large quantities of descriptive physical information, such as indicators of air and water quality, species counts, and area of forest cover. Typically, these informal accounting systems appear as national state-of-the-environment reports or in large physical environmental databases such as the STRESS system in Canada and similarly large databases maintained by several U.S. governmental agencies.
Physical accounting systems play an important role in accounting and policy formulation. They provide the underlying data for regulatory analysis and for development of the aggregates that underlie economic accounting. Moreover, they provide rich physical and intuitive measures of environmental impacts. At the same time, several factors complicate their use for policy purposes. First, the choice of appropriate physical units of measure is not obvious. Presumably, the units of measure should be relevant for some environmental-policy concern. Should a forest, for example, be measured in terms of its acreage, the volume of its timber, the variety of its biota (as evidenced by the number of available species), the stock of nontimber resources such as firewood and grasses, or the number of miles and acres of fishable waters? From the policy maker's point of view, the answer will depend on policy objectives: commercial timber management, firewood supply, recreational uses, erosion protection, species diversity, and so on. Additionally, when environmental assets have multiple uses, as in the case of forests, the units of the indicators are different (acres, cubic feet, number of species, cords of firewood, and miles of streams). The noncommensurate nature of the different attributes makes physical accounting rich in detail, but poor for making policy decisions and determining tradeoffs.
In all accounting systems, important questions relate to coverage, detail, and aggregation. In an effort to encompass the many policy issues involved, physical systems can easily become quite large and detailed. Of course, national accounting systems are also enormous data systems—but most of the vast data iceberg is under water, and only the monetary aggregates are visible in the published numbers. Indeed, large data systems are worth little for scorekeeping, modeling, or policy purposes unless they can be aggregated in such a way that they can be digested and understood. While economists often suggest that measures should be aggregated in terms of dollar values (or present values if there are streams of values over time), putting physical measures into a common unit of account often involves difficult valuation issues. In many cases (for example, protection of unique resources such as the wildness of Yellowstone or the visibility at the Grand Canyon), policy makers may be uncomfort-