From Bloomberg:
Stocks and commodities dropped in immediate reaction to a much larger-than-expected level of jobless claims, at 473,000 in the Feb. 13 week vs. expectations for 440,000. There are important special factors possibly affecting the data but their effects are unknown and the Labor Department isn't offering any explanations. There was extremely heavy weather through most of the nation in the reporting week, and results from four states had to be estimated including the key states of Texas and California with holiday backlog in the latter having skewed prior reports.
In addition, there is the WSJ:
The number of workers filing new claims for jobless benefits jumped by 31,000 to 473,000 in the week ended Feb. 13, the Labor Department said Thursday. Bad weather can make it harder to calculate jobless claims, among other things making it more difficult for state agencies that collect and process the numbers to adjust for seasonality and forcing some states to simply estimate claim levels.
But Marketwatch reported:
Bad weather around much of the eastern half of nation apparently had little net impact on new claims last week. Many state employment offices were closed but most people file by phone or online. However, it's likely that some people who lost work due to the storms filed for a claim, wrote John Ryding and Conrad DeQuadros of RDQ Economics.
I'll split the difference and say there was some effect, but it will take a few weeks to shake out. That being said, here is a chart of the data:
Notice that since roughly the end of November, claims have been stuck on a weekly basis between ~450,000 and ~475,000. This is a bothersome development. Before then we saw a nice, continually moving lower number. But now the number is stuck in a range. Some of this sticking is due to technical issues. This week's number is the second time in about 6-7 weeks there has been an issue related to collection of the data. That is responsible for some of the blip we saw. However, I'd like to see the weekly number start of move lower.
