The headline seasonally adjusted BLS job growth was below expectations. The internals say this is not a good report.

Analyst Opinion of the BLS Employment Situation

The household and establishment surveys were poorly sync’d this month. This was a crummy report – you know that when manufacturing and construction were the primary drivers in a service economy. Part of the explanation from Sentier Research:

The Census Bureau conducts nationally representative household surveys used to generate important statistics such as the monthly unemployment rate, housing characteristics, consumer expenditures, etc. All of these surveys are based on samples of housing unit addresses. Persons that have been displaced by Hurricane Harvey will no longer be at those addresses when the Census interviewers arrive or call, resulting in these households being listed as vacant. Without going into great detail these statistical series based on these surveys may be significantly affected even as the Census Bureau will make every effort to account for this disruption.

For the monthly Current Population Survey (CPS) used to measure employment and unemployment the monthly interviews are normally scheduled for the week of the month containing the 19th day and the questions refer to the previous week containing the 12th day.

The year-over-year rate of growth for employment was unchanged this month (blue bars on graph below). This is a year-over-year analysis which has no seasonality issues.

Economic intuitive sectors of employment were very soft.

This month’s report internals (comparing household to establishment data sets) was inconsistent with the household survey showing seasonally adjusted employment contracted 74,000 vs the headline establishment number of growing 156,000. The point here is that part of the headlines are from the household survey (such as the unemployment rate) and part is from the establishment survey (job growth). From a survey control point of view – the common element is jobs growth – and if they do not match, your confidence in either survey is diminished. [note that the household survey includes ALL jobs growth, not just non-farm).

The household survey added 77,000 people to the labor force.

A summary of the employment situation:

BLS reported: 156K (non-farm) and 165K (non-farm private). Unemployment rate degraded 0.1 % to 4.4 %.

ADP reported: 237K (non-farm private)

In Econintersect‘s August 2017 economic forecast released in late June, we estimated non-farm private payroll growth at 150,000 (based on economic potential) and 195,000 (fudged based on current overrun of economic potential).

The market expected (from Bloomberg / Econoday):

Seasonally Adjusted Data Consensus Range Consensus Actual Nonfarm Payrolls – M/M change 147,000 to 205,000 180,000 156,000 Unemployment Rate – Level 4.3 % to 4.4 % 4.3 % 4.4 % Private Payrolls – M/M change 140,000 to 200,000 177,000 165,000 Manufacturing Payrolls – M/M change -10,000 to 14,000 9,000 36,000 Participation Rate – level     62.9 % Average Hourly Earnings – M/M change 0.2 % to 0.4 % 0.2 % +0.1 % Average Hourly Earnings – Y/Y change 2.6 % to 2.7 % 2.6 % +2.5 % Av Workweek – All Employees

34.5 hrs to 34.5 hrs

34.5 hrs 34.4 hrs