ZeroHedge and EJ Antoni assert that yes, relying on the in-house series…
Figure 1: Total change in private employment (black bold), change attributable to government employment (blue bar), to private (tan bar). Source: BLS CPS, and author’s statistics.
If it seems unbelievable that government employment should fluctuate between positive and negative values, then you could be forgiven. A household survey is useful for many things, but finding month-to-month changes in sub-categories is not one of them. For example, compare the corresponding graph, using the data from the establishment study.
Figure 2: Total change in private employment (black bold), change attributable to government employment (blue bar), to private (tan bar). Source: BLS CES, and author’s statistics.
The household survey is well received as a major variable, as shown in Figure 3:
Figure 3: First difference of the log of CES nonfarm payroll employment (green), of CPS employment adjusted for nonfarm payroll employment (red). The NBER has defined recession days as shaded in gray. Source: BLS, NBER, and author’s statistics.
Likewise for government employment (in percentage terms):
Figure 4: First log difference of CES government employment (blue), of CPS government employment (red). The NBER has defined recession days as shaded in gray. Source: BLS, NBER, and author’s statistics.
This is why ZeroHedge was forced to turn to using CPS data, as well as CES not seasonally adjusted numbers. Wow.
Source link