Showing posts with label data analysis. Show all posts
Showing posts with label data analysis. Show all posts

Friday, May 31, 2013

Divergence in employment growth numbers

A few months ago I highlighted how looking at the employment data on a year over year basis provides a different perspective on the numbers.  Today I'd like to bring up a divergence between the private and public data.

yoy employment data and divergence them

The green and blue lines are the ADP and government year over year employment growth data, left scale and the red jagged line is the difference between their growth rates, right scale.  What's noticeable is the divergence between them.  While it would be better to show just the absolute difference between them regardless of sign, I'm not enough of a FRED graph Meister to figure that out right now.   Right now the >0.35% difference between them is the largest on record for this data series.

Is this divergence truly exceptional? It may have been greater in the past and only later revisions tightened up the spread.

I'm not making a prediction as to which is wrong, merely it's likely the spread will narrow in the future.

Monday, March 12, 2012

Revising employment data

In my last post I promised to look at how employment data is being revised over time. On Friday the latest employment data was released and previous data was revised upwards in both the seasonal and non-seasonal data series. While this graph below shows data from just last months revision I may start showing multiple threads to see how the revisions change over time once we have more data.

I know, not much data to look at yet. Hopefully we'll see something interesting as time progresses. 

Wednesday, February 8, 2012

This post (and employment data) will be revised

The recent employment report has been a great point of contention between the bulls, bears, and the black helicopter crowd. I won't rehash the report but there's something I mentioned in a previous posting about employment seasonality that caught my attention.

The beginning of a new year brings a large number of layoffs as seasonal workers for the holiday shopping rush are released.  As you can see from this first graph the non seasonally adjusted data and adjusted data vary greatly right now.
Total nonfarm payroll 
The difference is rather obvious to see but lets transform the data a bit and look at the difference between the normalized and raw data.


Subtracting one from the other exposes just how much the raw data can vary from the the adjusted data in an attempt to smooth out the seasonality of the data.  As can be seen we are at the peak of this delta and it will most likely be revised in the future due to the fact we are that noisiest point in the data collection period.  Will it be revised up or down? I haven't a clue, but I'm going to watch it and report to you how much these data series change over time.  Stay tuned.