I’ve been tracking the first-time jobless claims, and their revisions, since 1/1/11.
It sure looks like something is wrong at the Department of Labor. In the 73 weeks where we have both the reported number and the revised number, the reported number has been revised upwards 69 times. It has been revised downwards 3 times, and been unchanged once.
The last time there was a downward revision was 2/24/2011, over a year ago.
This means that, for 69 out of the last 73 weeks, the reported change has looked better than the actual change. Here’s how this works:
- In the weekly press release (example), the DOL compares this week’s reported number to last week’s revised number. I believe a better comparison would be this week’s reported to last week’s reported; and last week’s revised to the prior week’s revised.
- Revising the prior week’s number upwards reduces the apparent growth in weeks where there is a growth in first-time jobless claims; and increases the apparent drop when there is a drop. In some cases, the revision actually switches the dynamics, with a growth in reported-to-reported numbers showing as a drop in reported-to-revised.
- This reversal, where an increase is portrayed as a decrease, has happened 12 times in the 73-week period, and an astonishing 7 times in the 13 weeks since March 1, 2012. More than half the time since March 1, an increase has been reported as a drop in jobless claims.
The graph below shows the size of the revision by week. (The red line represents the average change.)
Now, what gets reported on the news every week is the DOL’s “headline” number: the change from last week’s revised number to this week’s reported number. This headline number has misrepresented the actual change, making the trend look better, every week since June 23, 2011.
Which means that the economy is not doing as well as is being reported, at least in first-time jobless claims.