Everyone Obsessing About Friday's Jobs Report Will Be Looking At This Chart On Wednesday

It's jobs week in America.  The main event is the June employment situation report, which will be released by the Bureau of Labor Statistics at 8:30 a.m. ET at Friday.

That's when we'll learn how many nonfarm payrolls were added by U.S. companies. (Economists are looking for 165,000 net new payrolls, with 175,000 net new private payrolls.)

However, on Wednesday morning we'll get the ADP employment survey, which many people consider to be a decent predictor of Friday's nonfarm payrolls number.  (Economists are looking for 160,000.)

And literally every economist has run their regressions on how accurate ADP is at predicting NFP. In fact, Deutsche Bank's Joe LaVorgna and High Frequency Economics' Jim O'Sullivan have created and circulated identical charts for everyone to ponder.

"Ignore ADP at your own peril," said LaVorgna.

Here's O'Sullivan (ADP est. 180,000):

The reliability of the ADP measure may have improved since Moody’s replaced Macroeconomic Advisers as the compiler of the data late last year, but that still leaves plenty of room for error. Last month’s miss was 40K, with the ADP series up 135K and private payrolls in the BLS report up 175K; total payrolls, including a 10K decline in the government sector, rose 165K. In the eight months since Moody’s took over, the ADP misses have ranged from -63K, with payrolls weaker than ADP, to +57K, with payrolls stronger than ADP. The miss without regard to sign has averaged 42K in the eight months since the change, down from 53K in the previous eight months.

High Frequency Economics

Here's LaVorgna (ADP est. 155,000):

In the chart below, we show the initially reported values for both private payrolls and the ADP survey. Over the last 12 months, the average error between the difference in private payrolls and ADP has been -14k, which is quite small since the standard error on private payrolls is about 75k per month. There does not appear to be a bias, as there have been six months when ADP has over- predicted private payrolls and six months when ADP has under-predicted private payrolls. The average forecast error between private payrolls and ADP without respect to sign has been 62k. Therefore, we would need to see a forecast miss in ADP larger than that for us to change our +145k nonfarm payroll estimate for Friday—private payrolls are projected at +155k.

Deutsche Bank

UBS's Kevin Cummins doesn't think much of the ADP's predictive power:

The ADP employment report has not been a useful real-time predictor of the change in private payrolls reported by the BLS. The ADP report started using a new methodology in October, although its track record is still largely unproven. The absolute value of the miss between the ADP and BLS data has averaged 42,000 since the change in methodology. For comparison, from January 2008 through September 2012 the average absolute miss was 81,000. On a month-to-month basis, the ADP data still appear to lack reliability as a predictor when the errors change direction.

UBS

The ADP jobs report comes out at 8:15 a.m. ET on Wednesday.  That'll be followed by weekly initial jobless claims at 8:30 a.m. ET.

View the original article here