Unlike labor data in general which economists deem to be a 'lagging indicator', that is the economy usually moves well in advance of job hires or fires, the weekly jobless claims data is considered to be a 'coincident indicator'; that is relatively real time. What has befuddled economists of late is the stubborn weekly jobless claims which are not falling below the 450K level, despite some gains (cough birth death model cough) in the monthly labor reports. Generally you need a figure below 400K to coincide with monthly gains but we have an anomaly. (which in my opinion is explained by the fiction that is monthly jobs statistics) I'd also assume with much more of the private workforce temporary - people working, then being let go, then getting a new assignment 9 months later, etc leads to a greater proportion of Americans to apply for unemployment more often; this is a structural change I believe is happening under the surface as just in time employment joins just in time inventory. Wax on. Wax off.
Whatever the reasoning, claims don't seem to be falling like you'd expect in a recovery. I am surprised myself since so much of the American workforce has been slashed the past few years ... I'd at least expect this number to be trailing off (dropping) from exhaustion. How much blood can you squeeze from the private sector stone? I don't want to read too much into any 1 week since the data is volatile (today we had a 472K print) and neither do economists, hence they used a simple smoothing technique - a 4 week average. Bespoke Investment Group has an interesting data set, charting the inverse of the 4 week average of jobless claims versus the S&P 500 the past decade. While not completely perfect (nothing is), it is actually quite an interesting correlation if not one with eagle eye precision.
What this chart also shows is how elevated the weekly claims are versus normal levels the past decade. We now find comfort in 425-450K claims a week as 'status quo' but aside from a few months in the 2001 recession this level we've settled at is worse than anytime in the decade. Which why it seems almost improbable it could roll over yet again and revisit the 500-550K range yet again, especially with 1/5th of the American workforce already unemployed or marginally attached.
(please note - I believe the large spike in claims in mid 2005 that did not result in the market dipping much was due to the job losses from Hurricane Katrina)