THE BREATHLESS talk about the miracle economy tends to have a techno-twang: wired this, wireless that, e-everything. But it turns out that one of the best things about the current prosperity -- the remarkably low unemployment rate -- has a decidedly low-tech origin. According to professors Lawrence Katz of Harvard and Alan Krueger of Stanford, the hidden hero is the temporary employment agency. And this holds a lesson for the Department of Labor.

The professors note that, as of last year, unemployment was 0.8 percentage points lower than it was in 1989, at the height of the previous business cycle. They observe that a quarter of that improvement comes from demographic shifts: Young workers tend to be in and out of jobs, and young workers make up a smaller percentage of the work force than they did 10 years ago. A further quarter of the improvement comes from the shocking expansion of prisons: Thousands of unskilled people who might have been unemployed are not counted as such because they have been locked away under Draconian mandatory sentencing laws.

What might have produced the other half of the reduction? Messrs Katz and Krueger consider the possibility that this could have come from government intervention. In 1993 Congress required states to provide job-search workshops, counseling and other help as part of the unemployment insurance system. Last year a million workers reported for these forms of aid. But the authors calculate that this reduced unemployment by 0.01 to 0.06 percentage points -- a tiny difference.

Then the professors deliver their punch line. Since the 1980s, the army of temporary employment agencies has grown markedly: Temp workers constituted 2.2 percent of the work force last year, up from 1.1 percent a decade ago. By making it easier for firms to find potential employees, temp agencies ensure that empty jobs are filled more rapidly. Moreover, the increased ease of new hiring discourages those with jobs from pushing wage demands as aggressively as they used to in flush times. Taken together, these two effects seem to account for an 0.4 percent fall in unemployment, or half the total improvement.

This conclusion underscores the familiar point that government is not the main influence on the health of the economy. It also suggests that Clintonesque micro-policies often get micro-results: The 1993 job-search law was not backed up with significant new funds to pay for extra training. But the Katz-Krueger research points to a further lesson. The private sector is increasingly good at placing skilled workers in jobs. Therefore, the Department of Labor should focus all the $5.3 billion in training resources on unskilled workers, who are unlikely to be helped by private temp agencies.

In the wake of the failed trade summit in Seattle, it will be tempting to reconstruct support for free trade by promising generous retraining grants to those who lose their jobs as a result of liberalization. The discomfort of laid-off steel or textile workers is real, and there may be a political case for aid. But the truth is that, so long as the strong economy persists, those workers can likely find new jobs. The same is not true for the long-term inner-city unemployed. They should be the priority for government.