Non-Farm Payrolls are an extremely volatile data series. Since the start of 2001, the standard deviation of the monthly change in NFP is an absurd 236,000. In other words, 68% of the NFP prints have been between +325,000 and -149,000, while the median is only 88,0000. That's a very, very big range. In terms of the difference month-over-month between changes in NFP, the median is 11,000, while the standard deviation is much lower at 134,000; in other words, we're not likely to see a swing from 288,000 last month to 62,000 this month. But the big standard deviation does mean that the data is really, really noisy. Intuitively, this makes sense. The U.S. economy is an enormously complicated machine, and measuring it as accurately as the Labor Department tries to (change in total payrolls to within 1,000 jobs) is, quite frankly, a thankless task. But the markets place a large emphasis on the report, so commentators do spend time trying to answer the question: "what happens to NFP this month?"
We prefer to use the initial release data, not the revised information that most of these data series rely on. The analysis below uses revised numbers, so it's not necessarily a great way to game the report, but it is a good indication of the drivers of the NFP. We took a look at using ADP, the previous month's NFP, the average of jobless claims within the reference month, jobless claims during the reference week, the average of continuing claims within the reference month, continuing claims within the reference week, percentage change in withholding taxes collected in a given month, and a seasonally adjusted measure of withholding taxes collected within a given month.
As illustrated above, the best indicator for NFP tends to be the ADP report. After that, the previous month's NFP (or, in other words, the momentum of NFP) is a good predictor. Jobless claims have a decent correlation to the report, while Continuing Claims are mediocre. Both the month-over-month change and seasonally adjusted month-over-month change in withholding taxes are a very weak indicator of NFP. So while withholdings for the current month are looking pretty weak (down 3.3% month-over-month or 1.8% after seasonal adjustment), we're not inclined to pay attention to that as an indicator.