Choosing a Welfare Indicator.
用斯里兰卡1981/82年消费金融调查数据，比较收入、总支出、食品支出、卡路里摄入和逆食品份额五种福利指标，发现指标选择会显著改变对储蓄、城乡贫困和营养不足的判断。
The choice of welfare indicator can make a substantial difference. Four examples drawn from the Sri Lanka Consumer Finance Survey 1981/82 will make this clear. First, when income is used as the welfare indicator, the bottom eight deciles of individuals are found to be dissaving, and when food expenditure is used, only the lowest decile is found to be dissaving. Secondly, 12.2 percent of the lowest 30 percent of individuals by total expenditure, but 20.6 percent of the lowest 30 percent by calorie intake, belong to the urban sector. Thirdly, the calorie intake of 46.5 percent of the population falls below 2,009 kilocalories per day (the requirement for the typical individual). By contrast, the income of only 22.9 percent of the population falls below the level at which an average calorie intake of 2,009 kilocalories per day is attained. Fourthly, the overlap between the lowest 30 percent of individuals by food expenditure and the lowest 30 percent by inverse foodshare is only 41.0 percent. In this paper we explore the reasons for, and the implications of, these differences. We examine five potential indicators of individual welfare: household income per capita (Y), household total expenditure per capita (T), household food expenditure per capita (F), household calorie intake per capita (K), and household inverse food-share (o-, defined as the ratio of total expenditure to food expenditure). We illustrate the discussion using the Sri Lanka Consumer Finance Survey 1981/82 (henceforth CFS). A detailed description and evaluation of these data can be found in Anand and Harris (1985, 1987).