The global economy is working. The rich may be acquiring an ever greater share of the world's wealth, the ecosystem may be collapsing, but - or so we believe - the poor are emerging from poverty. This is portrayed as the ultimate test of the great neo-liberal experiment: if, as the world's resources are privatised and its corporations deregulated, the war against poverty is being won, then the accompanying inequality and destruction can be accounted as little more than collateral damage.

There is only one set of figures which provides a global view of whether the incomes of the poor are rising or falling, and it is cited everywhere. The trend, it suggests, is slow but significant: between 1990 and 1999, the percentage of the world's people living in absolute poverty fell from 29% to 23%. Ugly as some of its characteristics may be, the existing economic model is helping the poor. The figures are compiled by the World Bank. It claims to know, to within the nearest 10,000, how many of the world's people are living below the international poverty line. The response of those who criticise the way the global economy works is to accept the bank's calculations, but to argue that there are more equitable and less destructive means of achieving the same results. But the figures are without foundation.

A new paper by the economist Sanjay Reddy and the philosopher Thomas Pogge demonstrates that the World Bank's methodology is so flawed that its calculations cannot possibly be correct. Not only does it appear wildly to underestimate the level of global poverty, but the downward trend it purports to show appears to be an artefact of the way in which it has been compiled. The World Bank's figures, against which the success or failure of the entire global economy is measured, are useless.

Most of the world's people do not use US dollars to purchase what they need, and a dollar's worth of currency in one part of the world can buy more than a dollar's worth in another. So to try to discover how many people live on less than the equivalent of $1.08 per day (deemed to be the absolute poverty line), the World Bank employs a method called "purchasing power parity". This measures the amount of goods or services which the equivalent of a dollar can buy in different countries.

The bank's calculations suffer, the paper suggests, from several fatal deficiencies. The most obvious of these is that its estimate of the purchasing power of the poor is based on the measure of their ability to buy any of the goods and services an economy has to offer: not only food, water and shelter but also airline tickets, pedicures and personal fitness training. The problem is that while basic goods are often more expensive in poor nations than they are in rich ones, services tend to be much cheaper, as the wages of the people providing them are lower.

If, for example, one dollar in the US can purchase either the same amount of staple foods that 30 rupees can buy in India, or the equivalent of 3 rupees' worth of services (such as cleaning, driving or hairdressing), then a purchasing power parity calculation which averages out these figures will suggest that someone in possession of 10 rupees in India has the same purchasing power as someone in possession of one dollar in America. But the extremely poor, of course, do not purchase the services of cleaners, drivers or hairdressers. A figure averaged across all the goods and services an economy can provide, rather than just those bought by the poor, makes the people at the bottom of the heap in this example appear to be three times as rich than they are.

The bank would derive a far more accurate view of the purchasing power of the poor if it measured only the cost of what they buy, rather than what richer people in the same economies buy. Complete figures do not yet exist, but Reddy's and Pogge's initial calculations, based on the cost of bread and cereals, suggest that the bank's analysis might have underestimated the number of the world's people living in absolute poverty by some 30%-40%.

As the service sector expands in poor nations, the bank's figures will create the impression that the purchasing power of the poor is increasing, whether or not their real economic circumstances have changed. The same false trend is established by a shift to the service sector in rich nations, as one dollar there will then buy a smaller proportion of the total of available goods and services. The relative purchasing power per dollar of the people of poor nations is increased by this measure, even though their absolute cost of living remains unchanged. When house prices boom in New York, the shanty-dwellers of Lusaka appear to get richer.

These statistical artefacts create a downward trend in the poverty figures where no real trend exists. The bank has exacerbated it by recalibrating the international poverty line to reflect the pattern of total global consumption. As the world economy migrates towards the service sector, the poorest people in the poorest nations appear to require less money than they might otherwise have needed to maintain their standard of living.

Perhaps more gravely still, the figures which appear to be so precise that we can tell to within the nearest 10,000 how many of the world's 6 billion people are suffering from extreme poverty are, in reality, based on a mixture of guesswork and wild extrapolation. The first of the bank's two principal surveys measured price levels in only 63 countries. Embarrassingly, China was not among them, and neither that nation nor India figured in the second survey (from which the trend has been established). A set of global poverty figures, presented with six-digit precision, which contains no useful comparative data from the two largest nations on earth, could be described as imaginative.

The bank's statistics, moreover, do not account for changes in inequality. If a nation's total consumption is rising only because the rich have become richer, the figures will not show this: they will suggest, instead, that everyone has prospered. Yet we know that in many countries - especially those in which the privatisation, deregulation and reduction in social spending introduced by the neo-liberal model have been most extensive - the rich are becoming richer at the expense of the poor.

That the key global economic statistic has for so long been derived by means which are patently useless is a telling indication of how little the men who run the world care about the impact of their policies. If they cannot be bothered even to produce a meaningful measure of global poverty, we have no reason to believe their claim that they wish to address it. Development on earth proceeds at present without any reliable means of determining whether or not it is making the poorest people poorer.