Of Nowcasts and Revisions | Econbrowser
Don’t pay too much attention to the headline change in employment.
The accounts of the employment release focused on the net change in April (above consensus), but more nuanced assessments noted the downward revisions. The following two graphs highlight the distinction between growth rates and levels. The first graph shows differences.
Figure 1: Month-on-month growth rate of private nonfarm payroll employment, April release (bold black), March release (dark blue), Bloomberg consensus of 5/5 (pink square), and author’s nowcast based on ADP (sky blue), all calculated as log differences. Source: BLS via FRED, Bloomberg, author’s calculations.
The actual change in private NFP growth exceeded the Bloomberg consensus shown relative to unrevised March. In this sense, the announcement surprised. (My nowcast based on ADP numbers far exceeded actual.). The actual m/m change was 0.17%, compared to 0.12% Bloomberg, and 0.27% in the ADP-based nowcast.
The previous months were revised down (as shown in the black line). In levels, the picture is quite different. The second graph:
Figure 2: Private nonfarm payroll employment, April release (bold black), March release (dark blue), Bloomberg consensus of 5/5 (pink square), and author’s nowcast based on ADP (sky blue) +/- two standard errors (gray lines). Source: BLS via FRED, Bloomberg, author’s calculations.
Figure 2 demonstrates that, despite the upside surprise on the April change in private NFP, the actual level is lower than the Bloomberg consensus implied (assuming no prior month revisions – we actually don’t know what the forecasters thought about this point).
The ADP based nowcast is off by 0.206% (274 thousand), of which 0.105 (139 thousand) are accounted for by revisions.
Clearly, net job creation is slowing, with 3 month trailing moving average is down to 182K/mo from 271K/mo at January.