RARELY does an economic indicator provide as much fodder for politicians and pundits as the unemployment rate. Far more than, say, current accounts or capacity utilisation, unemployment is something everyone can understand: you are either in work, not in work, or looking for work. As such, it is easily seized upon as an indicator of the broader health of an economy, or even of workers' eagerness to revolt.

The issue of unemployment has loomed especially large in America in recent months. That is partly because there are presidential elections in November, and much will hinge on whether George Bush can convince voters that an apparently booming economy is producing jobs. A glance at the unemployment rate would seem to give him the answer he wants. The unemployment rate has fallen from a post-recession peak of 6.3% in June to 5.6% last month, though that is still higher than the 5.0% that many economists consider to be the “natural rate” of unemployment—one that results merely from the normal or “frictional” patterns of job gains and losses at any one time.

But the unemployment rate is, in fact, a poor measure of economic health. It is defined as the fraction of the people in the labour force—those who are actively seeking work and available for it—who cannot find a job. And it relies on surveys to determine who is, in fact, actively seeking work rather than enjoying a spot of leisure. It is that subjectivity that makes the unemployment rate such a flawed statistic. A better question by far is how many people are employed—ie, are being paid by someone for doing something, since this should be less subject to doubt.

Or so you might have thought. Yet there has been a fierce debate in America recently over even this humble statistic. That is because the number employed in America is also still measured using surveys, and the two that are widely used tell different stories. One is taken of over 400,000 firms with formal payrolls. Another asks 60,000 households whether people in them are working. But both are hostage to the usual limitations of using small samples to estimate employment for the whole economy, though obviously to different degrees. They are, moreover, subject to big revisions. And both have their advantages.

The payroll survey uses a bigger, more easily verifiable sample. On the other hand, the household survey may better capture a rise in jobs among new small businesses and the self-employed, both of which seem to have accounted for a lot of new employment in the recent recovery. According to the household measure, nearly 139m Americans were in work in January, even more than had jobs at the height of the boom in March 2000. By the payroll measure, some 130m were in work—a fall of nearly 2% since employment peaked.

Left-leaning pundits naturally prefer the payroll survey. The Bush administration and its friends prefer the household version. Still, even the latter's figures would make job growth in the current economic recovery anaemic by historical standards.

Concerns over employment data are not just an American problem. According to a recent report from Barclays Capital, Germany's employment statistics may be overstating the numbers of self-employed because of a government initiative to subsidise previously unemployed workers in starting their own business. Combined with other shenanigans, this may produce an army of “hidden unemployed” of 1.4m, estimates the report, some 30% more than the number of officially unemployed. In Japan, the unemployment rate has never risen above 5.5% in recent years, despite a decade-long economic funk. That is in part because firms are reluctant to sack workers for social reasons.

Flawed though they may be, the employment numbers are of fundamental importance. Two crucial questions for economic output and for the suffering caused by unemployment are: what portion of the working-age population does not work and how many of those that do not work want to do so?

Regardless of which survey you believe, more people of working age are at work in America than in Europe. America's employment rate is just over 70%—almost ten percentage points higher than Europe's. In other words, less than a third of working-age Americans are not in work, whereas in Europe the figure is closer to 40%, though the gap between the two economies has been closing in recent years, as America's employment rate has fallen and Europe's has risen.

Many of those that do not work would almost certainly like to. By the OECD's reckoning, the ranks of those who could be mobilised are thus far bigger than those that are formally classed as unemployed. Indeed, in most countries, according to the OECD, there are far more gains to be had in bringing inactive workers into work than in reducing unemployment to its “natural” rate. In Italy, for example, the OECD reckons that more than a fifth of the working-age population could be brought into work, and some 17% in Spain and Greece.

In the euro area, the relatively lower employment rate explains much of the region's lower GDP per person. And low employment is often the fault of misguided policies that discourage people from working, such as high payroll taxes; marginal income taxes that penalise the work of a lower-paid spouse; rules that make sacking workers expensive; and generous benefits that encourage the work-shy to be classed as disabled, to name but a few.

Such structural problems play a huge role in the differences in the wealth of nations. The trouble is that fixing them can be politically fraught. Just ask Gerhard Schröder, Germany's chancellor, who resigned this month as head of his party, because of resistance to a package of modest reforms. Having jobs is one thing; working quite another.

Copyright © 2004 The Economist Newspaper and The Economist Group. All rights reserved.