ADP: A Modest Gain For September Payrolls

Private non-farm payrolls increased by 162,000 last month, according to this morning’s ADP Employment update for September. That’s down a bit from August’s revised 189,000 gain, which suggests that we should keep our expectations in check for Friday’s official September jobs report from the Labor Department. Nonetheless, there’s nothing conspicuously dark in today’s data dump to suggest that the slow growth trend rolled over last month. In fact, when you consider today’s ADP release with yesterday’s mild rebound in the ISM Manufacturing Index last month, the case is a bit stronger for expecting September to remain in the growth camp (once all the month’s numbers are published).

Here’s how the last 12 months of ADP data compare with the Labor Department estimates:

How should we interpret today’s ADP report in terms of guidance for Friday’s government release? The short answer: cautiously. There are times when the ADP estimate implies that its Labor Department counterpart will look relatively upbeat. For example, when ADP’s July 2012 estimate was published on August 1, the odds looked favorable (odds that turned out to be accurate) for anticipating a solid improvement in the government’s estimate due for release two days later vs. its monthly predecessor. Why? As I discussed at the time when the ADP number hit the street, the monthly divergence between the ADP and government estimates was near a three-year high.
Why’s that relevant? It starts by assuming that the ADP and Labor Department numbers will converge through time. That’s a reasonable assumption, given the high-quality work behind the ADP data, which comes from Macroeconomic Advisers. The empirical record also tells us to expect that the two employment series will track one another fairly closely generally. The month-to-month estimates, however, are quite noisy vs. one another. As a result, when the noise becomes unusually loud, so to speak, the odds for mean reversion are relatively high, as was the case for anticipating the July Labor Department figures.
Mean reversion tends to prevail eventually when it comes to the differences between the two data series. As the chart below shows, the monthly gaps in the ADP numbers vs. the Labor Department data tend to fluctuate randomly around zero. That’s a strong sign that the two estimates won’t vary too far for too long. Indeed, the average monthly difference between the two data sets for the past 10 years is a mere +2,000, or the statistical equivalent of zero in context with monthly changes that can swing up or down by 200,000 or more.

Although the divergence through August is climbing again, it’s still below the highs of the last several years. We can’t rule out the possibility that Friday’s employment report will deliver some improvement over July’s tepid 103,000 advance in private payrolls. But the second chart above suggests that expecting a sharply better jobs report in two days doesn’t look like a high-probability event (as it did two months ago).
Anything’s possible, of course, and sometimes big surprises catch the crowd unaware, despite the best-laid plans of statistical analysts. But if we’re dealing in probabilities by looking at the numbers available, today’s data tells us to expect more of the same for Friday’s update: slow growth.