Unlike broad-based tariffs that are negotiated through the WTO or previously through the General Agreement on Tariffs and Trade, subsidies, anti-dumping duties, quotas and non-tariff barriers are generally produced by a domestic political process. This not only allows non-economic factors to dominate the discussion, it is also a seed-bed of corruption.

Extreme free traders also ignore the effect of tariffs in raising tax revenue. Of course, the economy is optimized if tariffs are zero, but it is equally optimized if tax rates are zero, and we are generally moving further away from rather than approaching that desirable state of affairs. Just as a tariff discriminates between domestic producers and importers, causing fewer goods to be imported than would be the case at a zero tariff rate, so an income tax discriminates between work and leisure, causing fewer goods to be produced than at a zero tax rate.

In the small-government societies before 1914, tariffs were often the major source of revenue for government. In late 19th-century America, the post-Civil War Republicans set tariff rates so high that the government generated a permanent excess of revenue, paying off Civil War debt and then accumulating a cash surplus. The 1880 Democratic presidential primary was fought on the slogan "Tariff for Revenue Only"--not for protectionism beyond the necessary level to fund government--producing the immortal Thomas Nast cartoon of their befuddled presidential candidate Winfield Scott Hancock inquiring "Who is Tariff, and why is he for Revenue Only?" Speaker "Czar" Thomas B. Reed, of the 1888 GOP Congress proclaimed in turn, "God help the Surplus." He then proceeded to spend it, an attitude not unknown in more recent years.

A "level playing field" tax system (of course, an unattainable ideal) that did not discriminate between consumption and production, nor between imports and exports, would have a tariff level lower than the income tax rate because income taxes are imposed on profits while tariffs are imposed on revenues. Since profit margins vary from item to item, a "level playing field" tariff would, in theory, be levied at a varied rate, with higher rates on high-margin items such as semiconductors and telecom equipment, and lower rates on low-margin items such as textiles. In this way, tariff rates on imports would be equilibrated approximately with taxes on incomes. All economic activity would be taxed to approximately the same extent.

This is not to advocate still a further increase in the size of government. Such a rational tariff system, to the extent it raised revenue, would allow for tax reductions in other areas. In the United States, taxes could usefully be reduced on corporate dividends, which remain subject to significant double taxation at overall rates that still exceed 50 percent, and on modest self-employment incomes, which, including doubled social security contributions, are also subject to high marginal tax rates.

In summary, a rational tariff system would allow modest protectionist tariffs to guard against exchange rate and weather driven disruption, but would not permit high tariffs, quotas, subsidies, non-tariff barriers or discrimination between sources of supply. It would allow for cuts in other government taxes. And it would need a credible world body policing it.

To see how such a system would work in practice, look at the three trade items that caused the most discord at Cancun: steel, textiles and agriculture. In a rational trade system, the United States and European steel industry would benefit from a low tariff of no more than 10 percent, which would provide American industry with modest assistance in periods such as early 2002 when the dollar was inordinately strong. However, the long-term lack of competitiveness of Big Steel (as distinct from the mini-mills) against cheaper labor and more efficient Asian competition would cause painful but necessary restructuring and downsizing in this industry. Costs of past mistakes, such as the excessive pension provisions in past union contracts, would not be subsidized by the American buyer of steel products. Whether they should be subsidized by the U.S. taxpayer as a way of easing the frictional costs of competition is outside the scope of this article (though I would be against doing so).

Similarly, U.S. and west European textile and garment industries are structurally uncompetitive with Asian imports because of their high labor costs. Specialty manufacturers, particularly producers of high-fashion apparel, can survive, but others must outsource most of the manufacturing process to cheap labor countries. (Sewing is particularly labor intensive, while cutting is largely automatable.) A modest overall tariff at the 10 percent level will not significantly slow the long-overdue migration of this industry to the Third World; the problem here is structural, not exchange-rate related. The Third World is right to be outraged by import quotas in textiles and by the inordinate delay to 2005 in removing them, as agreed by the Uruguay trade round in 1994.

In agriculture, most price fluctuations are caused neither by exchange rate movements nor relative labor costs, but by Mother Nature. Hence, in societies such as the United States where farming is highly capital intensive, governments must protect their farmers' investments against poor harvests or, more damagingly, against good harvests in other parts of the world that produce temporary gluts of produce and prices too low to cover the farmer's fixed costs. In the United States, Europe and Japan, the solution to this since the 1930s has been to subsidize the farmer, often paying him for acreage taken out of production or dumping exports on other countries at subsidized prices. This is incredibly wasteful, an unfair barrier to imports from the Third World and destabilizing to the incomes of Third World farmers. The objectives of farm subsidies (other than simply diverting wealth to the pockets of agribusiness) can be achieved much more elegantly by a mechanism that has been wrongly derided since it was abolished in 1846: the 1815 British Corn Laws, or rather their 1827-28 modification.

The central principle of the Corn Laws was that treatment of corn imports should vary with the market price of corn. Under George Canning's proposed version of 1827, the duty (paid by the foreign exporter) rose in rough proportion to the drop in the corn price and fell with the rise in price, with corn entering Britain free of duty when prices rose to high levels in times of scarcity. Farmers received protection against gluts and consequent drops in price, while consumers were protected against bad harvests. There were no subsidies, and the net cost to government was simply that of administration. In three respects, therefore, its lower cost to government, the lack of distortion to the domestic market (which rewarded the most efficient producers accordingly) and the lack of incentives to dump produce on the world market, the Corn Laws system was greatly superior to modern agriculture subsidies.

A modern Corn Law would apply to agricultural products in general and would work in the same way. It would replace subsidies, quotas (as in sugar) and above all export bounties, thus preventing the huge economic waste of over-production followed by export dumping (illegal under WTO rules in any sector other than agriculture). Ideally, it would be applied at equal levels in the United States, the EU and Japan.

The mechanism for negotiating trade agreements also needs to be modified. The current WTO, with its reliance on one country, one vote decision-making and a requirement for unanimity is, as demonstrated at Cancun, hopelessly open to subversion by anti-trade fanatics playing on the fears of the less sophisticated countries of the Third World. Equally, there are too many deals "stitched up" in advance of full WTO meetings between powerful trading blocs, which also tend to lock in negotiating positions and prevent progress being made.

A central difficulty in the WTO is its voting mechanism, which is based on a flawed analogy with democracy, and should instead be based much more soundly on a shareholder-type voting structure, with voting power proportional to the amount of trade a country enjoys. For example, voting rights could be allocated every five years according to the sum of a country's imports and exports of goods and services in the previous quinquennium. Countries with a great deal of trade in proportion to their GDP, such as Singapore, would be disproportionately weighted in the WTO, reflecting their disproportionate importance in the world trading system, whereas countries such as Brazil that maintained high trade barriers would be less heavily weighted. There would be a natural institutional bias toward free trade, proper in such a body.

Having established a voting system that truly represented the importance of each country in world trade, there would then be no need for unanimity. But a supermajority--maybe 75 percent--of votes would be required to pass tariff agreements. It could be that blocs of countries such as the EU would attempt to get together and form a blocking minority. But there is an obvious riposte. If they acted as a single entity in trade negotiations--as did the EU, for instance, at Cancun through Pascal Lamy--then their intra-regional trade would cease to be counted in determining their weighting, just as trade between Iowa and Kansas would not be used in determining the U.S. weighting. In addition, the problem of NGOs determining the atavistic reactions of small and unsophisticated Third World countries would go away. Such countries would represent only a small portion of world trade and so would have only a small WTO vote.