How is the non-farm payroll tracking the most complete data?

The BLS released Quarterly Census of Employment and Wages (QCEW) data for the second quarter of 2013 this morning. Since it's based on the unemployment insurance system's records, it offers near-universal coverage of the job market (as opposed to the monthly payroll survey, which is just a sample, though a very large one). It shows that total employment grew 1.6% in the year ending in June, compared with 1.7% reported by the monthly payroll numbers. This is a reversal of the last release, which had the QCEW growing 0.1 point more than the payroll survey. But an 0.1 point difference in either direction is trivial, and confirms that the payroll survey is doing a good job at tracking this weird recovery/expansion.

Earnings figures in the QCEW are difficult to compare with the monthly payroll numbers, since the definitions and coverage are different, and the series is not seasonally adjusted. They showed a 6.9% decline in wages from the first quarter to the second, following a 1.1% decline from 2012Q4 to 2013Q1. Quarterly declines are normal in the first and second quarters in the QCEW earnings series, but the second quarter decline is larger than normal (the first quarter was quite close to average).

Finally, the number of employing establishments rose 0.7% for the year ending in the second quarter, a decline from the 2.0% gain in the first. The trend, however, is working its way slowly higher after the sharp (and unprecedented since the series began in 1976) decline in establishments during the Great Recession. The pace of startups is well below the 2-3% average of the 1980s and 1990s, and even the 1.7-1.8% average of the mid-2000s. Establishment formation is extremely important to employment growth, since young firms are the primary creators of new jobs. So this is good news for continued employment gains, but offers little hope for an acceleration in the pace of job growth over the next year or two.