Posted by: Josh Lehner | March 7, 2012

Effectiveness of the ADP Report, Update

The following provides a quick update on the previous, extended post that examines the effectiveness of the ADP employment report. That post used data through June 2011 to examine the differences between the official BLS employment data and the sneak peek that the ADP report provides a couple of days in advance. The following uses data through January 2012 and accesses the effectiveness in the same manner. How have things changed in the past 7 months? Not really that much, however there is one slight difference.

At the time of the previous post, the average difference between the two reports was only 200 – that was the BLS report, on average, would be only 200 jobs different (larger) than what the ADP report stated. However, while this is true over the entire period (see how closely the two lines move together in the first graph below) there is quite a bit of noise on a month to month basis. Extending the data through January, and including revisions to both the ADP and BLS data since then, the average is now 2,200 jobs, however the month to month noise remains large. One standard deviation of the differences in the reports is 76,000, thus two standard deviations is 152,000. That means the with 95% confidence, one can state that the BLS report will show employment changes equal to the ADP report changes, plus or minus 152,000.

As with a lot of economic data, these indicators tend to be noisy, the key is to take a step back and examine longer term trends and patterns; not get hung up on any one particular data point. Clearly the ADP and BLS employment reports move together over time and in about the same magnitude, however in any particular month the differences can be nearly 200,000.

A simple regression between the two series yields the following results. The goodness of fit and coefficient results are nearly identical to the previous version from last year, as expected. (Meaning that the past 7 months have not diverged wildly from the previous 11 years worth of data)

The last graph illustrates the above equation’s fit (the top two lines) and residuals (the bottom blue line along with two standard deviation dotted lines). One can see the clear overall pattern however the month to month noise (residuals bouncing around) is evident as well.

Click here for the ADP Employment website (click Historical Data in the right column for an excel spreadsheet).


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

%d bloggers like this: