THE BLOG
01/21/2014 05:50 pm ET Updated Mar 23, 2014

Somebody Got the December Jobs Data Very Wrong -- And It Matters to Your Money

Early each month a lot is riding on the jobs and unemployment reports announced by the BLS and corporate surveys (the most prominent being ADP's). Billions upon billions of dollars are made and lost in the stock and bond markets based on those reports. Vital Federal Reserve monetary policy decisions are directly linked to which way the numbers turn and Congress looks to these data as it deliberates issues like extension of unemployment benefits or minimum wage levels. CNBC even has a contest to "nail the number" -- but this month, nobody came close -- the winner turned out to be over a third off the mark!

The ADP survey of private sector job creation in December 2013, announced January 8, reported a net gain of 238,000 jobs in that month, data which is very consistent with recent monthly job report trends from both ADP and the government's Bureau of Labor Statistics (BLS). Yet on Friday, January 10, the Bureau reported a net gain of 74,000 in U.S. jobs, reflecting only 87,000 in the private sector (while governments cut 13,000). Is anybody a winner here when the numbers are so disconnected?

We have known for some time that the only thing we really know about the BLS's first estimate of job creation is that it is wrong. The variability factor cited by many economists is a range of 80,000 up or down, meaning that we may well have created over 150,000 net jobs in December, or actually lost 6,000! The Bureau itself expresses a "90 percent confidence" factor in its data, but only within a 90,000 plus or minus range.

The BLS has already revised its November 2013 report upwards by 38,000, and has made similar adjustments routinely throughout the year and in previous years Indeed, the BLS's initial job creation number for June and July of 2013 were eventually revised down by a total of 74,000 out of an initially reported 350,000 -- a 21 percent decrease. This correction, announced in early September, may have played an important role in the Federal Reserve's mid-September decision to delay its "taper" of QE3 bond-buying stimulus program until December. Fortunately, they got the correct information on time, but just barely. When will we know the real numbers for December and what will they run out to be? The ADP survey, which had been tracking closer to the BLS report in recent months until December, may give us a clue.

Each survey is broken down into various sector categories, which do not have exactly equivalent definitions, but are close enough for "back-of-the-envelope" comparisons. The most intriguing maybe the disparity between the BLS's report that showed a net contraction in construction employment of 16,000, while the ADP survey reported a net gain in construction of 48,000. This is an unusually wide differential in the monthly figures. Next would be the spread between professional and business services job growth of 53,000 in the ADP report, while the BLS totaled only 19,000 net, reflecting what it estimated as a net decrease in accounting positions of 25,000, a very unusually large decline not reflected in any recent reports of layoffs from the Big Four accounting firms.

Again, ADP showed a stronger December rise in manufacturing job growth of 19,000, compared with only 9,000 in the BLS survey. The BLS survey also showed little job growth in transportation and logistics from the previous month, while ADP reported a strong 47,000 net gain in its trade, transport and utilities category.

Such disparities to the downside in the BLS estimate as compared with the ADP survey require a deeper look under the hood. The BLS survey is based on a survey of payrolls of 145,000 employers, and is focused on the count of individuals on the payroll specifically during the workweek of the month that includes the date of the 12th. In December, that turned out to then be the coldest week of the quarter, unusually so in the western US. The BLS itself reported in its separate household survey of the unemployment rate that 273,000 workers temporarily "left" the workforce during that same week due to weather-related causes and thus may not have been on the payrolls counted for the job creation survey.

That level of weather-related employment reduction is far above what would be the normal December pattern of around 100,000, and the BLS job survey's seasonal adjustments are unlikely to have accounted for this anomaly accurately the first time around. If just half the weather-related job "cuts" during the cold week of December 12, were added back, the December jobs report could have shown a total growth of 210,000 jobs, much closer to the ADP total of 238,000, and much more in line with consensus expectations of 195,000 in net employment gains for the month.

ADP also uses the payroll data from the week including the 12th of the month, from 411,000 employers, which represent an aggregate of 20 percent of total US employment. This is a far larger sample base than used by the BLS. In fact, ADP revised and expanded its outreach just this fall with the express intent of more closely aligning its findings with the BLS, and for two months that's how it worked out. But not December.

The key question in terms of discerning which of the December competing employment reports of the BLS and ADP is actually closer to the true situation would seem to turn on which survey had the better sense of how many of those not on the payroll the week of December 12th were really just temporary, weather-related layoffs from shopping malls, theaters, construction sites and other commercial and governmental operations that were forced to shut down or cut back because of the early wintry conditions. It appears that the BLS may have been trapped by its own methodology into treating such job 'losses" as though they were permanent and ADP somehow eluded that trap.

Look for a big upward revision eventually (by March if not February) in the BLS job growth data for December. We'll just have to wonder what the stock and bond markets would have done on January 10 if the initial December BLS report had been more in line with ADP and economists forecasts. Meanwhile, we can take some comfort from the predictions of a slight thaw in the cold weather for the week of January 12th!