It's the Wednesday before the jobs report, and that means one thing: listening to mid-90s pop and considering what charts to make for Labor Department numbers I haven't seen yet.
But in addition to that, it means a fresh ADP jobs count. This month, the payroll processing firm said America's private firms added 238,000 jobs in December, well above consensus expectations of around 200,000. But it's also not much of a predictor of what Friday's payrolls number will be.
ADP's figures are watched closely and are often considered bellwethers for upcoming jobs reports. For example, today's report boosted 10-year Treasury yields to levels not seen since 2011.
It's true that the ADP figures have done an excellent job during the long run of tracking the government's figures.
To put it in statistics language, the r for these two series – "r" being a common measure of correlation – equals roughly 0.93. On a scale of -1 to +1 (with both positive and negative 1 signaling perfect correlation and zero meaning exactly zero correlation), 0.93 is a very tight relationship.
But look at the numbers during a shorter time period and the relationship looks much more jagged.
It isn't just the chart that tells a story here; starting at 2010, r equals something closer to 0.57 – only a moderate correlation. And looking at just the last 12 months, r is less than 0.3. Put into non-math language: the shorter the timeframe, the less the two figures have to do with each other, meaning that it's hard to use any given month's ADP count as a predictor of what the government's jobs report will say.
During the last 12 months, for example, the ADP and government job counts have missed each other by an average of nearly 43,000 jobs. That's the difference between a so-so jobs report – 160,000 jobs added – and a relatively strong one – 203,000 jobs.
Of course, it wouldn't make sense for the two figures to be in perfect lockstep. For one thing, ADP uses data from the large sample of employers to whom it provides services; the government performs surveys of U.S. employers. In addition, the Labor Department includes public-sector jobs in its count; ADP only looks at private firms.
Likewise, the government's data isn't going to be perfect – it will go through two more monthly revisions after this as more data come in (and then still further revisions after that).
None of this is to say the ADP report doesn't give useful information. One key thing the ADP report provides that the government does not is a breakdown by firm size, telling how small business hiring fared compared to large firms.
In addition, as Bill McBride noted on his Calculated Risk blog today, the latest ADP report "suggests employment growth above expectations." So while the report may predict the magnitude of job growth, it at least can signal the direction.