Unfortunately, the evidence is not particularly persuasive. Even the World Bank estimates that removing rich countries' barriers to developing country exports would generate only very small income gains for the latter. When fully removed by 2015, developing countries as a group might experience a gross domestic product increase of 0.6 per cent. Almost all the gains would be concentrated in fewer than a dozen developing countries, and many would actually lose from rich country trade liberalisation in key sectors. The ending of the multi-fibre agreement early this year, for example, has hurt textile exporters across the developing world as super-efficient producers in China gobble up the market.

Moreover, the removal of producer or export subsidies in the west hurts countries that consume large quantities of the subsidised exports. Removing developing countries' own trade barriers may indeed help their exports. But the estimates of large income gains come from trade models that audaciously omit some significant costs and exaggerate the net gains. One such cost is the loss of tariff revenue, which often amounts to 10-20 per cent of government revenue. The revenue has to be made up by alternative taxes (such as sales or income tax), which have their own "distortionary" impact. Another cost omitted by the trade models is the handicap that lack of protection imposes on an infant industry sector. If the theory of comparative advantage worked, people and capital "released" from existing businesses would be re-employed in other, more "efficient" activities. But the theory assumes full employment, and therefore no significant transition costs. It simply assumes the problem away.