More than half a million unemployed people say their fortunes improved dramatically last month: They got a job.
Now if only someone could prove it.
When the government surveyed U.S. households in July for its regular employment report, it found 629,000 people who said they had just started work. But when the government asked companies how many jobs they had added to their payrolls, the answer was 32,000.
If they're not working in a store, office or factory, what are those 597,000 other folks doing? Working as consultants? Selling bric-a-brac on eBay? Mowing their neighbors' lawns?
Or are they actually unemployed but so ashamed that they're lying about it?
"I can't tell you," said Tom Nardone, chief of the Labor Department's Division of Labor Force Statistics. "We just don't know why there's a difference between the surveys."
These new workers resemble the dead in the movie "The Sixth Sense": Only some people can see them.
Those catching a glimpse are mostly Republicans. Vice President Dick Cheney, for instance, can see them clearly. They're free-lancers, private contractors, people working at home. They're not on the roster of any corporation's human-resources department but are prospering anyway.
They're people like his wife.
At an Aug. 11 campaign appearance in Missouri, the vice president said Lynne Cheney "does very well in terms of her own professional career and line of work, but she doesn't work for anybody. ... If you're in business for yourself, if you've got your own small business and so forth, you don't get picked up by those other numbers."
Calculating employment is a massive task. To estimate payroll levels, the Labor Department's Bureau of Labor Statistics queries 400,000 so-called "worksites" every month about their hiring activities.
For whatever reasons, "worksites" haven't been in a hiring mode for a long time.
But when the government asks 60,000 people directly about employment, as it also does every month, the jobs picture looks healthier.
Although the 629,000 increase in July was unusually high, the cumulative increase in the household survey since March 2001 is 1.8 million jobs.
"The divergence between the household and the payrolls survey is very striking, and I'll leave it to statisticians to try to reconcile those numbers," Treasury Secretary John Snow said after the July numbers were released earlier this month. "I suppose that the real number lies somewhere in between."
Nardone, the Labor Department statistician, has been trying just such a reconciliation. By smoothing, revising and adjusting the July data, he eliminated one-third of the new employed, leaving 402,000 unexplained new workers.
But over the longer term, tweaking the household survey to bring it in line with the payroll survey actually increased the gap. Since July 2003, the cumulative difference, even after reconciliation, is 1 million jobs.
Commentators and economists with conservative affiliations have been echoing Snow's remarks, when they're not exceeding them.
"The BLS for more than a decade has been undercounting job creation, unable to keep up with changes in the structure of American business," columnist Robert Novak charged recently.
Allan Meltzer, a Carnegie Mellon economist and a scholar at the American Enterprise Institute, a conservative think tank, said in an interview that the terrific household numbers might overstate what is really going on, but that "the truth is much better" than the weak payroll numbers, which signal a troubled economy.
The summer's lousy payroll numbers, Meltzer said, "just don't fit very well with what we're seeing -- rising wages, increasing disposable income, improvements in manufacturing employment."
Even partisans acknowledge that the household numbers, which are used to calculate the monthly unemployment rate, are highly variable. While the payroll numbers have increased every month for the past year, if sometimes very modestly, there have been three occasions when the household survey reported a net decrease in jobs.
In February, for instance, when payrolls rose 21,000, household employment plunged 265,000. No one in the administration suggested then that the real employment truth lay somewhere between the two.
Many economists reject the notion that the monthly household reports offer insights that the payroll survey doesn't. The administration "is just trying to muddy the waters," said Dean Baker of the left-leaning Center for Economic and Policy Research. "The household numbers are just too erratic."
If you focus on the self-employed, the matter becomes even murkier.
The number of full-time self-employed workers increased in July by 175,000, which lends initial credence to Cheney's claims that many of the newly employed are working for themselves.
However, very few of the 9.5 million people who are now self-employed exist at Lynne Cheney's level. A Bush/Cheney campaign spokesman declined a request to put a dollar figure on her husband's assertion that she is doing "very well."
Lynne Cheney earned money last year from Reader's Digest, whose board she served on, and the American Enterprise Institute. The couple said the $321,000 they donated to charity last year came "primarily" from royalties on her books.
Many self-employed writers are closer to charity cases themselves. Consider Diane Feen, a 54-year-old fashion writer in Boca Raton, Fla. "My take-home pay for the entire summer will be about $1,700. I see my income and work load dwindling every year," she said.
To make ends meet, Feen lives with her mother. "I think that many people are working from home because they have no choice. My single friends are struggling more than ever," she said. "And I used to be connected and upper-middle-class. I used to shop at Bergdorf and Saks.
"And I used to buy goat cheese."