Data Review

February 18th, 2016 9:06 am | by John Jansen |

Via Stephen Stanley at Amherst Pierpont Securities:

Initial unemployment claims surprisingly fell further in the week ended February 13 to 262,000, the lowest reading since October.  My read based on the advance state-by-state data is that the snowstorm-related jump in the number of new filers two weeks ago was still unwinding, as some of the affected states (Pennsylvania, Tennessee, New Jersey, and New York) posted sizable declines for a second straight week.  Based on that presumption, it would make sense to focus on the average over the past three weeks, which is 272K, i.e. very close to where I expect the underlying trend to emerge once the seasonal noise dissipates.  Next week’s release will cover the period including Presidents Day, so we are probably another 2 weeks from getting a clean read, but the turn-of-the-year window during which claims are noisy should be winding down.  In any case, while I do not take the 262K reading at face value, there is no reason to believe that the underlying trend has backed up even a little bit from the extraordinarily low prevailing range seen in the second half of 2015.

Meanwhile, the Philadelphia Fed manufacturing survey index was essentially unchanged in February at -2.8 (vs. -3.5 in January).  Most key components were modestly in negative territory.  The driving force as well as the outlier appears to be the inventories measure, which deteriorated further to -17.1, the lowest reading in almost 3 years.  I believe that we are closer to the end than to the beginning of the inventory correction, but it is clearly still ongoing.  This is dragging down new orders and shipments (among other things).  Nonetheless, yesterday’s IP release provides some hope and, as noted above, I do think that we are most of the way through the correction, so conditions in the factory sector, while they will probably not be good at any point this year, should improve somewhat as the year progresses.

Be Sociable, Share!

Post a Comment