Seasonal factors are most likely still distorting the reported unemployment claims numbers. The problem lies with auto plant layoffs, which have come earlier this year than is usual — they usually happen in July, a period I've highlighted in the first chart, which shows unadjusted claims.
[Click all to enlarge]
With auto layoffs coming at a time when they weren't expected, the seasonal adjustment factors treat that as if it were a genuine rise in the number of unemployment claims, and that's why we got a spike in the adjusted data (bottom chart) in April and May. But there will be payback for this mistake come July, if I'm right.
When the actual number of layoffs doesn't rise as much as the seasonals expect, the adjusted numbers will show a surprising decline; something similar happened last year, when adjusted claims experienced a sudden dip in July. So in a few weeks we should expect to see some "unexpected" declines in the number of unemployment claims, and that will likely come as a positive shock to the market. Unfortunately, it's all much ado about nothing.
I note also that unadjusted claims have been relatively low and stable for the past few months. There haven't been any big things happening in the real world.
As this last chart shows, on an unadjusted basis the number of people collecting unemployment insurance has been declining steadily all year. The labor market is slowly improving, not deteriorating. The progress is disappointingly slow, but it's progress nonetheless.