MOST people who have a view on the matter—regardless of whether they are critics of globalisation or advocates—accept that global inequality is getting worse. Most official agencies either say or seem to suppose that global inequality is rising. In a new paper* on the subject, Xavier Sala-i-Martin of Columbia University quotes the typical and widely cited United Nations' Human Development Report. In 1999 this said:

In 1960, the 20% of the world's people in the richest countries had 30 times the income of the poorest 20%. In 1997, 74 times as much. This continues the trend of nearly two centuries. Some have predicted convergence, but the past decade has shown increasing concentration of income among people, corporations and countries.

How does the UN know that global inequality has grown much worse? First, its economists say, inequality has worsened within countries. Second, inequality has worsened across countries. From these two things, the UN reckons, it follows, third, that inequality among the people of the world is rising as well.

Mr Sala-i-Martin agrees that inequality has probably increased, on average, within countries. The picture is not clear-cut, however. Inequality has gone up in some countries and down in others. (Rapid globalisation does not push all one way: emerging-market globalisers such as South Korea and Indonesia have seen inequality fall.) On the whole, though, the author reckons that within-country inequality has most likely gone up during recent decades.

What about inequality across countries? On this, the UN neglects an important point. If you measure incomes in terms of purchasing power rather than at market exchange rates, incomes are a lot more equal. (The reason is that the cost of living is lower in poor countries.) When the UN says that the incomes of the richest 20% were 30 times bigger than the incomes of the poorest 20% in 1960, and 74 times bigger in 1997, it is using market exchange rates. In purchasing-power terms, the corresponding ratios were 11 and 15. Despite the fall after 1980 (when the ratio was 16) the trend for the period as a whole is nonetheless up.

Yet another measure (the cross-country variance of income per head) confirms this. Over the past 30 years, rich countries have grown richer and most of the very poorest have stayed very poor. This is the pattern that Harvard's Lant Pritchett referred to in the title of his well-known paper, “Divergence, Big Time”.

But hang on. Given propositions one and two—rising within-country inequality, and rising across-country inequality—it does not follow, as you might suppose, that global inequality itself is rising. Why not?

Mr Sala-i-Martin explains. Imagine that five-sixths of the world's population live in poor and stagnant economies, and one-sixth in rich fast-growing ones. In across-country terms, this gives you “divergence, big time”. Now imagine that one poor but very populous economy starts to grow very quickly. At the same time, inequality within this country worsens somewhat.

Despite its size, this country is only one data-point in the across-country comparisons: its rapid growth is not enough to make any difference to divergence. So you have rising within-country inequality and rising across-country inequality. Yet one-sixth of the world's population, by assumption, is seeing its incomes rise rapidly towards those of the rich. Inequality measured across all the people of the world, therefore, may very well be falling.

A far-fetched case? No, Mr Sala-i-Martin points out, this is exactly what has been happening. The big poor country growing very fast is China. (India has also been doing pretty well.) If you simply weight the across-country measures of divergence by population, you see not a rising trend of inequality, but the opposite: as the author puts it, not “divergence, big time” but “convergence, period”.

Country variance weighted by population may be better than the unweighted sort, but still it ignores inequality within countries. Mr Sala-i-Martin therefore sets about combining both kinds of information, to see how income is distributed across the world's people. He finds that rising global inequality is “nowhere to be seen”. This is true on seven different measures: “the Gini coefficient, the variance of log income, two Atkinson's indexes, and three generalised entropy indexes”. So there.

The headcount of poverty worldwide has fallen by some 400m

For those more interested in relieving poverty than in narrowing the gaps between rich and poor, the results from the estimated distribution are equally pleasing: the proportion of people living on less than a dollar a day fell from 20% in 1970 to 5% in 1998; the proportion living on less than two dollars a day fell from 44% to 8%. The headcount of poverty worldwide has fallen by some 400m.

The only bad news is that, after the respite provided recently by surging globalisation, inequality may well resume its long-term historical trend and start rising again in due course. The reason is that China and India will no longer be poor—and if the world's poorest countries, mainly in Africa, continue to stagnate, the global dispersion of incomes will widen. Whether the main problem here is African poverty or global inequality (caused by China and India leaving poverty behind) is one for the UN's economists to think about.

*"Mr Sala-i-Martin's study is "The Disturbing 'Rise' of Global Income Inequality", NBER Working Paper No 8904. It is available on the NBER website. A version also appears on Mr Sala-i-Martin's own website.

"Divergence, Big Time" by Lant Pritchett was in the Journal of Economic Perspectives, Summer 1997, Vol 11, No 3. An earlier version appeared as World Bank Research Working Paper No 1522.

Copyright © 2002 The Economist Newspaper and The Economist Group. All rights reserved.