FIXING NON FARM PAYROLL DATA ERRORS
BLS Mistakes Happen
Sometimes the BLS methodology misfires and payroll shifts from one month into another. The total payroll doesn't change, just the timing. The Updated NFP data eliminates those distortions in order to reveal the true payroll activity.
Fixing, Not Second-Guessing
Updating is very rare and only targets events that are (1) publicly acknowledged and (2) significant (~100K+ payroll impact). An example is the Verizon strike in August when ~45K workers went on strike and then returned to work in September. It was a non-event: specific to Verizon and completely reversed. It also significantly distorted the payroll picture and needs to be addressed.
Reversing the strike shifts payrolls as follows:
- August payrolls: increased 45K (45K strikers no longer counted as a drop in payroll)
- September payrolls: decreased 90K (45K returning strikers no longer counted as new payroll PLUS October's now higher payrolls reduced September's growth by 45K)
- October payrolls: increased 45K (September's payroll figure dropped 45K)
Instead of a September surge and October fall back, as the chart below shows, payrolls moved up slowly and steadily.
For 2011, there are two meaningful corrections.
- January blizzards: Per the February BLS report, back-to-back blizzards blew out their methodology. Impact: 100K January jobs were counted as 100K extra February jobs.
- Verizon strike in August
Trends Revealed
Updating the NFP data uncovers important trends
1. Q2 payroll collapse is more evident: businesses dramatically and immediately reigned in expenses
2. Q3/Q4 payroll growth has been more steady and positive