cent of the value of crop production—the same percentage as annual production increases. In other words, these estimates suggest that current increases in farm output in Indonesia's uplands are being achieved almost wholly at the expense of decreases in future output. Because the upland population is unlikely to be smaller in the future than it is now, soil erosion represents a transfer of wealth from the future to the present. By ignoring the future costs of soil erosion, the sectoral income accounts significantly overstate the growth of agricultural income in Java's highlands.
In recent years, a number of researchers have struggled to define sustainable agriculture. Most of these definitions encompass elements of agricultural productivity maintenance, farm profitability, and reduction of environmental impacts, but they have been qualitative, not quantitative. Also, most definitions of agricultural sustainability have failed to incorporate productivity of the natural resource base when calculating agricultural productivity. The notion of agricultural sustainability has therefore been of considerable conceptual utility but only limited operational usefulness to policy makers and researchers attempting to determine how various policies and technologies affect agricultural resources.
An NRA framework differs from conventional financial and economic accounting in some significant ways (Faeth, 1993). In conventional accounting, the financial value (net farm income) of a production program to farmers takes into account current and future transfer receipts but ignores environmental costs. Using the NRA framework, net farm income is defined to include the value of changes in soil productivity, the farmer's principal natural asset. This definition is consistent with business and economic accounting standards, which incorporate asset formation and depreciation in measures of income. By contrast, the same farming technique's economic value to society (net economic value) includes environmental costs that farmers' activities impose on others, such as damages related to surface water, but ignores transfer payments.
Tables 3 and 4 present examples of this NRA methodology. The tables compare net farm income and net economic value per acre for a predominantly corn-soybean rotation in Pennsylvania, with and without allowances for natural resource depreciation. Column 2 of Table 3 shows a conventional financial analysis of net farm income per acre per year. The gross operating margin ($75) (crop sales less variable production costs) is shown in the first row. Because conventional analyses make no allowance for natural resource depletion, the gross margin and net farm operating income are the same. Government subsidies ($16) are added to obtain net income ($91).