Here is a chart showing the 21-day moving-average of Federal withholding-tax collections, compliments of the distinguished gentlemen over at DailyJobsUpdate.com.

As you can see, the average has dropped into the $5.5 billion range – the lowest level in a year!

KABOOM!!!

(click chart to enlarge):

Ha, ha! Gotcha! Hope you liked my “Tyler Durden” impression! Just a little “NFP Eve” jobs humor!

If you know what’s wrong with the interpretation above, you get a gold star. Tell your mom that I said it was OK.

Of course, tax revenues always drop this time of year. The next chart is the same as above, but from one year ago:

Almost identical, right? However, the upper chart has not been adjusted for this year’s big social-security tax cut. So, it would be substantially higher if it were not for that. And that makes sense because the jobs that the economy has added over the past year are still there. Maybe we are tipping over into recession, but there is no sign of that yet in the withholding-tax data.

Do you think that it’s just a coincidence that September is the worst month for the stock market? Or could this seasonal slow-down in economic activity influence the market?

Note: It is possible to make such moving-average charts of the withholding-tax data because it is reported every day, unlike other economic data that is only reported weekly, monthly, or quarterly. The charts above, and more, are updated every day at DailyJobsUpdate.com.

On Sunday, Zero Hedge ran a piece by Nicholas Colas of BNY in which Colas stated that federal withholding-tax receipts were down 4% year-over-year for the month of February (through the 22nd). And he drew the exact wrong conclusion. February is in the books now and receipts were actually up 3.8%. See for yourself. First, look at the final total from February 2010 (click image to enlarge):

Inside the red box, you see $139,589, right? Now, let’s look at the final report for February 2011 that was released today:

Inside the red box, you see $144,891. And that is $5,302 higher than $139,589. And since the numbers are expressed in millions, that’s a $5.3 billion beat.

That is flat-out year-over-year growth of 3.8%.

Colas didn’t even get the sign of the number right! Ridiculous!

And there was a huge tax-cut too, right? The only rational, objective way to interpret this data is that the economy has been creating jobs at a very healthy clip. If you like, you can change “creating” into “re-instating jobs lost during the Great Recession”, but the idea that the economy is contracting, stalling, or even slowing down just doesn’t fly. As a matter of fact, things are likely accelerating, though we can’t tell for sure because the tax-cut has clouded this data.

So why did Colas get it so horribly wrong? He made a couple of rookie mistakes. First of all, he used a period that was too brief. Paychecks go out every other Friday, monthly, or twice-monthly. When comparing two periods of withholding data, you absolutely must use periods that contain equivalent pay-cycles.

Colas’s second rookie mistake was to synchronize to the calendar rather than the work week. February 2010 began on a Monday, while February 2011 began on a Tuesday. That will often make a big difference.

So, Colas didn’t use enough data, and the periods he compared were not equivalent. He calculated the right number, but for non-comparable data. Here is what Colas did:

Here is the correct way to handle this sensitive data:

Notice how I grouped the data into work-weeks (#5 through #8 of the year). Also notice that I made sure to have a complete pay-cycle.

That’s how it’s done.

Going by work-weeks, we come up with a gain of 1.71%. And I believe that is the correct interpretation. Calendar February was up 3.8%, but that should be taken with a grain of salt because the calendar was a bit out of sync.

In any case, the fact that the data is higher after a large tax-cut is pretty miraculous.

As to the markets, they may well continue to react to events in the Middle East rather than to our expanding economy.

Earlier in the year, a few people asked me about using social security and medicare taxes, as opposed to withholding-tax data, to evaluate national labor trends. And now I have issued an edict. In summary, it’s bad data. Social Security and Medicare revenue are calculated from tax forms using methods determined by law. The withholding data is counted-up from actual money deposited into the Treasury Department’s bank account by employers, and is therefore much more pristine.