top of all this comes the effect of basing estimates on national accounts rather than on household surveys It is revealing to consider why according to the Chen-Ravallion study poverty fell relatively slowly on their household-survey measure. It was not because of an increase in within-country inequality-in other words, brisk growth in average consumption was not being hogged by the better off. It was because growth in average consumption was slower than what growth in national incomes, as measured in the national accounts, would lead you to expect Growth in consumption per head across the countries in the Chen-Ravallion sample was less than 1% a year between 1987 and 1998, according to the household surveys. Growth in consumption per head according to the national accounts was more than 3% a year Mr Deaton notes that a plethora of new data has so far failed to resolve this issue" because the new sources are mutually contradictory". Summing up, he states: If the surveys are wrong and the national accounts right either inequality has been widening in ways that our data do not appear to show or poverty has been falling more rapidly than shown by the dollar-a-day counts. If the surveys are right there has been less growth in the world in the 1990s than we are used to thinki It would be a mistake to presume that either source of data is better in principle. Surveys are famously prone to error because of bad or fluctuating design, discrepancies in samples and them prone to understate the consumption of the poor, but also to overstate the growth ore poor execution. But national accounts have drawbacks as well especially in poor countries. For instance, they fail to capture some sorts of non-market income and consumption. This make consumption of the poor as incomes rise and as more activities fall within the scope of market ransactions Take your pick Still, most of the discrepancy between the survey estimates and the national-accounts estimates-with the surveys persistently pessimistic on trends in poverty -is probably due to the fact that as people get better off, they are less likely to respond (accurately or at all)to surveys. As a result, as countries get richer, the ratio of survey consumption"to "national accounts consumption is usually found to fall. Consistent with this, the ratio of the two measures is highest in the poorest countries Mr Deaton argues that both sources ought to be used, though combining them properly raises a host of difficult technical issues. Meanwhile, the truth about global poverty and inequality presumably lies somewhere between the extremes suggested by the two methodologies One can at least conclude that the official World Bank data, used by the n and other agencies are too pessimistic: poverty has most likely fallen faster than these widely cited figures suggest, and possibly fast enough to reduce the global headcount of those living on less than a dollar a day even as population rises. More accurate answers will require more work to be done. In the meantime, however, the official position on global poverty ought to start, at minimum, to acknowledge the uncertainty surrounding the figures and further, to concede that the truth is likely to be better than the official figures say Beside the point?top of all this comes the effect of basing estimates on national accounts rather than on household surveys. It is revealing to consider why, according to the Chen-Ravallion study, poverty fell relatively slowly on their household-survey measure. It was not because of an increase in within-country inequality—in other words, brisk growth in average consumption was not being hogged by the better off. It was because growth in average consumption was slower than what growth in national incomes, as measured in the national accounts, would lead you to expect. Growth in consumption per head across the countries in the Chen-Ravallion sample was less than 1% a year between 1987 and 1998, according to the household surveys. Growth in consumption per head according to the national accounts was more than 3% a year. Mr Deaton notes that a plethora of new data has so far failed to resolve this issue “because the new sources are mutually contradictory”. Summing up, he states: “If the surveys are wrong, and the national accounts right, either inequality has been widening in ways that our data do not appear to show, or poverty has been falling more rapidly than shown by the dollar-a-day counts. If the surveys are right, there has been less growth in the world in the 1990s than we are used to thinking.” It would be a mistake to presume that either source of data is better in principle. Surveys are famously prone to error because of bad or fluctuating design, discrepancies in samples and poor execution. But national accounts have drawbacks as well, especially in poor countries. For instance, they fail to capture some sorts of non-market income and consumption. This makes them prone to understate the consumption of the poor, but also to overstate the growth of consumption of the poor as incomes rise and as more activities fall within the scope of market transactions. Take your pick Still, most of the discrepancy between the survey estimates and the national-accounts estimates—with the surveys persistently pessimistic on trends in poverty—is probably due to the fact that as people get better off, they are less likely to respond (accurately, or at all) to surveys. As a result, as countries get richer, the ratio of “survey consumption” to “national￾accounts consumption” is usually found to fall. Consistent with this, the ratio of the two measures is highest in the poorest countries. Mr Deaton argues that both sources ought to be used, though combining them properly raises a host of difficult technical issues. Meanwhile, the truth about global poverty and inequality presumably lies somewhere between the extremes suggested by the two methodologies. One can at least conclude that the official World Bank data, used by the UN and other agencies, are too pessimistic: poverty has most likely fallen faster than these widely cited figures suggest, and possibly fast enough to reduce the global headcount of those living on less than a dollar a day, even as population rises. More accurate answers will require more work to be done. In the meantime, however, the official position on global poverty ought to start, at a minimum, to acknowledge the uncertainty surrounding the figures and, further, to concede that the truth is likely to be better than the official figures say. Beside the point?