There is lots of talk about the “fiscal cliff” the U.S. faces at year end, as stimulus and tax cuts go away.

So the last thing the government needs now is market distrust in its job numbers. But, as analysts dig into the government job numbers, questions are increasingly being raised about the reliability of the data, from questionable revisions in the weekly jobless numbers to the odd changes in unemployment rates.

For 59 out of the last 60 weeks, the weekly jobless numbers have been revised, after the fact, always in the same direction: higher. That’s unheard of.

Those revisions higher make the present week’s unemployment number look better in comparison, more so since the markets often treat the prior week’s revision as an afterthought.

And there is statistical manipulation in the unemployment rate, too. The government’s reported unemployment number doesn’t include people who stopped looking for work, but who want jobs.