None of the three major wire services covering today's report from the Department of Labor on initial unemployment claims is reporting the major news: For the first time in a long while, actual claims filed during the most recent week ended January 12 were almost 6 percent higher than the number filed during last year's comparable week, an indication that the current employment market may be worse than it was a year ago. Instead, all three are headlining how today's questionably created seasonally adjusted claims number is the lowest in five years.
Both weeks had five business days. Both weeks represented the first such week in the new year. So how did higher raw claims result in the lowest seasonally adjusted claims number in five years, a number which is 8 percent lower than last year's comparable week? The answer, as will be seen after the jump, is that the seasonal adjustment factor used this year is sharply higher than the one used last year.
The following graphic lays it out:
DOL's data in the top section of the graphic show that actual claims (NSA, or not seasonally adjusted, in the green boxes) this year exceeded last year's analogous week by over 30,000, or 5.8%. I believe that this is the first time that year-over-year claims in truly comparable full business weeks have risen in two years, if not longer. Yet seasonally adjusted claims (SA, in the red boxes) came in 8 percent, or 29,000, lower than last year.
The three calculations below DOL's data show why this happened. The 2013 seasonal adjustment factor of 1.660 (expressed as "166.0" in results obtainable at this interactive DOL link) is far higher than the factor of 1.442 (or "144.2") used in last year's analogous week. Again, both weeks had five business days, and both weeks represented the first full business week after the new year. Why this year's seasonal adjustment factor is 15% higher (1.660 divided by 1.442) than last year's is something I certainly can't explain, and I suspect that DOL can't defensibly explain it either. Readers won't see anything except vagueness in what DOL had to say in the wire reports which follow below.
The highlighted column on the right shows that if last year's seasonal adjustment factor had been applied to this year's raw claims, seasonally adjusted claims would have come in at 385,000, or 50,000 higher than the "five-year low" DOL is reporting today. I believe that 385,000 more properly reflects underlying labor market conditions, which are not improving, and that today's reported 335,000 does not.
The Associated Press, Bloomberg News, and Reuters all hemmed and hawed about seasonal adjustments, but failed to report the really important news that raw claims increased (links are to screen grabs at roughly 9:30 a.m., as the wires typically revise their reports as the day goes on; bolds are mine):
Associated Press, via Christopher Rugaber ("US JOBLESS AID APPLICATIONS FALL TO 5-YEAR LOW")
The number of Americans seeking unemployment aid plummeted to a five-year low last week, a hopeful sign the job market may be improving. But much of the decline reflects seasonal volatility in the data.
... The department seasonally adjusts the numbers to account for such trends, but the data can still be choppy.
Bloomberg, via Alex Kowalski ("Jobless Claims in U.S. Fell to Lowest Level in Five Years")
The number of Americans filing first-time claims for unemployment insurance payments fell more than forecast last week to the lowest level in five years, pointing to further improvement in the labor market.
... A spokesman for the agency said the drop may reflect the difficulty the government has in adjusting the data after the holidays when seasonal workers are let go.
Fewer claims indicate businesses have grown comfortable with their current headcounts, a necessary development before hiring starts to pick up.
Reuters, via Lucia Mutikani ("Jobless claims drop to five-year low")
The number of Americans filing new claims for unemployment benefits tumbled to a five-year low last week, a hopeful sign for the sluggish labor market.
... While last week's decline ended four straight weeks of increases, it is probably not the start of a new trend or a sign of a material shift in labor market conditions as claims tend to be very volatile around this time of the year.
This is because of large swings in the model used by the department to iron out seasonal fluctuations.
While it would be fair to say that AP and Reuters restrained their enthusiasm a bit in their content, the same cannot be said about their unqualified headlines. Bloomberg's Kowalski was especially week in mentioning "falling claims" in his content when actual claims really rose.
Of course it's the headlines which will dominate radio and TV broadcasts, and it's the headlines people will see in news feeds on their computers, tablets, and smart phones. They all convey a sense of improvement which has no support in reality.
Cross-posted at BizzyBlog.com.