the marginal damages caused by an additional ton emitted in 1987. Kilowatt hours generated per unit of emissions, a single-factor measure of environmental productivity, improved much more rapidly than capital, labor, or energy productivity between 1970 and 1987. The index of emissions fell by 30 percent over this period, while electricity output increased. Had emissions per kilowatt hour remained at the 1970 level, the electric power industry would have emitted 10 million more tons of sulfur dioxide in 1987 than it did.
If the electric utility produces both kilowatt hours and emissions, a MP measure can be constructed incorporating the changing output mix. The share of emissions in total output should be measured in economic terms using their ''shadow prices" (i.e., estimates of the marginal damages to the economy occasioned by an additional ton of each pollutant). Such estimates are hard to obtain, because damages vary across the country and are not reflected in market transactions. As the best available approximations, we used estimates prepared by U.S. Environmental Protection Agency and reviewed by the Office of Management and Budget. Their best estimate of marginal damage costs for sulfur dioxide emissions in 1987 is $637 per ton, with a range of $290 to $1,612. Damages to health, materials, agriculture, and visibility are included in the total, but contributions to acidic deposition or climate change are not. Analogous numbers for particulate matter and nitrogen oxides are $2,550 and $230 per ton, respectively.5
Using these shadow prices to estimate the value of emissions in each year,6 the cost of emissions as unpriced outputs of the electric power industry in the 1980s was about as large as the cost of labor to the industry. Hence, emissions have as great a weight as labor in a generalized productivity index.
In summary, productivity growth in the private electric power industry during these years appears two to three times as high if its progress in reducing economically damaging emissions is taken into account. In addition, productivity in the industry increased more rapidly in the 1970s, when emissions were being reduced more rapidly, than it did in the 1980s, when the rate of decline was more modest. This is contrary to the conclusions of conventional MP measurements, which do not incorporate environmental imports (Table 5).
Although this is a preliminary exploration, it suggests that technologies that reduce environmental damages contribute significantly to economic productivity. They do not merely raise production costs. The example also suggests that it is important to measure environmental dimensions of productivity to avoid one-sided assessments.
Economic indicators and analyses that count the cost of environmental protection but ignore the cost of environmental degradation and the loss of natural assets mislead both public and private decision makers. The problems with productivity measurement, for example, have led to serious misunderstandings about