How not to save American jobs.

Mr. Borjas is professor of public policy at the John F. Kennedy School of Government, Harvard University. This article draws upon work reported in "How Much Do Immigration and Trade Affect Labor Market Outcomes?" written jointly with Richard B. Freeman and Lawrence F. Katz and published in the Brookings Papers on Economic Activity, 1997.

THE defeat of "fast track" boosts protectionism in both political parties. The labor unions and Dick Gephardt were already emotionally committed to "fair trade" as a means of saving American jobs. They will now calculate that it is also good politics. Pat Buchanan will be similarly tempted to stress tariffs rather than immigration reform in his "America First" package. So the political debate may continue to ignore the potential economic trade-offs between free trade and immigration -- although new research shows they can be dramatic.

Trade and immigration have risen sharply in recent decades. In 1970, exports and imports stood at 8 per cent of GDP; by 1996, they had risen to about 19 per cent. Much of this increase is trade with the Third World: by 1996, nearly 40 per cent of all imports came from there. At the same time, there has been a huge increase in immigration, mainly of less-skilled workers.

The economic arguments for free trade and "open" immigration are superficially identical. When we import garments made in Southeast Asia, some workers in the American garment industry either suffer wage cuts or lose their jobs. These losses, however, are more than offset by the benefits accruing to consumers in the form of lower prices. In short, although some sectors might lose from trade, the economy as a whole gains. Similarly when we import immigrant workers into the economy, the additional competition in the labor market lowers native wages. At the same time, native-owned firms gain because they can hire workers at lower wages, and consumers gain through lower prices. As with foreign trade, the gains accruing to those who consume immigrant services exceed the losses suffered by native workers.

Finally, both trade and immigration increase the "effective" labor supply of particular groups of American workers. When we import a car from Japan, we effectively import, say, 150 man-hours of unskilled labor, 230 man-hours of engineering expertise, and so on. Like immigration, this increase in labor supply increases economic pressures on workers with the same skills and qualifications as those used to produce the imports.

When it comes to "saving American jobs," however, what matters is how trade and immigration respectively shift the effective U.S. labor supply. It is well known, for instance, that immigration has substantially increased the supply of workers who are high-school dropouts. But how has foreign trade changed the labor supply?

Workers in import industries tend to be less educated, those in export industries, well educated. Put simply, imports hurt the less skilled and exports help the skilled. As a result, even balanced trade -- where the dollar value of goods exported equals the dollar value of goods imported -- may have a sizable distributional impact, shifting income from the less to the more skilled.

Once we recognize this, we can estimate the impact of trade on income distribution. This calculation would construct, by skill group, the man-hour equivalent of each car imported, say, and each copy of software exported. We can then calculate the "factor content" of trade -- the implicit number of workers involved -- by adding up these man-hours over all imports and exports.

Such calculations reveal that the economic impacts of immigration and trade are surprisingly different. In particular, immigration changes the labor supply to a much greater extent than trade. The immigration that occurred between 1980 and 1995, for example, increased the number of high-school dropouts in the U.S. by 21 per cent; trade increased the effective labor supply of these workers by only 4 per cent. At the other end of the skill distribution, immigration increased the supply of workers with a high-school diploma by only 4 per cent; trade increased it even less -- by less than 1 per cent.

Because trade and immigration have such different effects on labor supply, they must also have different effects on income distribution. Between 1980 and 1995, the wage gap between high-school dropouts and high-school graduates increased by 11 percentage points. The impact of immigration on the relative supply of high-school dropouts accounts for almost half of this decline in the relative wage. Trade accounts for only 10 per cent of the decline -- one-fifth of the immigration impact.

A PROVISO should be added. If there were no immigration, we would want to import many goods now produced by immigrants. In that case, the effects of trade on wages might be larger.

Even so, immigration probably has a much larger impact than trade over the long haul. If the U.S. were to stop trading, trade would cease to have an impact on the effective labor supply of less-skilled workers. The goods imported in the past decade have long since been consumed, and trade leaves no footprint. Immigration, however, is a permanent increase in the labor supply. Even if the U.S. were to stop admitting immigrants tomorrow, the impact of past immigration would continue throughout the immigrants' working life -- and beyond, as their children and grandchildren entered the labor market.

In the end, when it comes to "saving American jobs" (and much else), immigration is a much bigger deal than trade. Multiculturalists like Dick Gephardt can hardly be expected to recognize this; Pat Buchanan's blindness is harder to explain.