Everybody is worried about the nation’s dismal employment situation, and that worry has prompted news organizations, pundits, market watchers, and others to focus intently upon any and all economic metrics that gauge the problem. On the first Friday of every month, the non-farm payroll report from the Bureau of Labor Statistics is hotly anticipated, and can move the markets for days afterwards.
As a more immediate measure of employment, news organizations have begun focusing on the initial unemployment claims reported every week by the Department of Labor. This comes out every Thursday, and usually prompts reports such as this, from Bloomberg:
The number of Americans filing first- time claims for unemployment insurance payments was little changed last week, adding to evidence the labor market recovery may have paused.
Applications for jobless benefits decreased by 2,000 to 370,000 in the week ended May 19 from a revised 372,000 the prior week, Labor Department figures showed today. The initial claims matched the median estimate in a Bloomberg News survey of economists. The number of people on unemployment benefit rolls and those receiving extended payments dropped.
However, is this really a decent measure of how the job market is performing? Maybe not.
It is worth noting that the government applies very wonky, hard to follow layers of seasonal adjustments to the raw data. The point is to smooth out changes in the labor market that do not have to do with the economy. But how good are these adjustments?
Check out this graph, which tracks the adjusted initial claims since the start of the recession. The dotted line is something I added to give a sense of the general movement:
Notice that the drops in the initial claims seem to happen in spurts, with a lengthy period of stagnation thereafter. Also, notice that these drops and stalls seem to occur at the same time every year. The drops begin late in the year and end early in the next year, while the stalls generally go from the late winter into the mid-fall.
That suggests to me that these seasonal adjustments are not that great. In other words when journalists report improvements in the labor market, per the weekly initial claims, they might just be picking up on the quirks of the seasonal adjustments.
There is actually some positive evidence to back up this claim. Fortunately, the government also provides the raw data, which has not had the seasonal adjustments applied. And we can apply our own kind of seasonal adjustment by looking at the change between one week’s reading and the reading from the same week last year. That is a good apples-to-apples comparison of how jobless claims are moving over time.
The following chart does that by tracking the percentage increase or decrease in weekly claims from a year ago.
Notice that the rate of growth in claims plummets pretty quickly through most of 2009, and by the end of that year we actually start seeing year-over-year decreases.
But then the slope of the line flattens out entirely, so that for the last two years, the unadjusted claims in any given week have typically been 10 percent fewer than they were a year ago.
In other words, the apparent swings in the seasonally adjusted claims entirely disappear. What we see is a steady unwinding of claims – good, but very far from great.
The lesson from all this? Some economic indicators are better than others. The weekly claims, in my opinion, do not give us a real sense of how the labor market is performing. Over the last two years we have seen periods of strength and weakness in terms of jobs added, but the unemployment claims have actually been steady.
So, I would not pay a whole lot of attention to this number every week.
Jay Cost is a staff writer for THE WEEKLY STANDARD and the author of Spoiled Rotten: How the Politics of Patronage Corrupted the Once Noble Democratic Party and Now Threatens the American Republic, available now wherever books are sold.
