sures. In the designing of environmental indicators, policy issues should dictate the choice of variables and the focus of the research. Measures should focus on human health and on social, economic, and ecosystem effects, rather than simply on pollutant concentrations or similar intermediate variables (see recommendation 4.3).

The panel considered the issue of the frequency of estimation and publication of natural-resource and environmental accounts. Because the underlying physical activities generally take place at a slow pace, particularly relative to business cycles, it is not sensible to aim for reporting in the satellite accounts more frequently than on an annual basis. Annual reporting is recommended for those activities—particularly subsoil assets and forests—for which annual data are readily available. For other measures, including input-output analysis, measures of comprehensive or sustainable income, and similarly complex constructions, quinquennial reports may be a reasonable goal. Frequent analysis and reporting are not necessary given the source data, costs, and temporal evolution of assets and activities that are being measured. Neither the data nor the likely uses of such accounts would suggest the need for monthly or quarterly data, particularly given the problems created by the short-run volatility of mineral commodity prices.

5.10 The panel recommends regular periodic accounting in the natural-resource, environmental, and other augmented accounts. The accounts for subsoil assets and forests could be developed, calculated, and reported on an annual basis. For other measures, less frequent accounts, perhaps quinquennial, would be appropriate and cost-effective.