Does seasonal adjustment of jobless claims unnecessarily complicate and distort the numbers?

Question

Bloomberg.com says about the latest jobless claims:

"First-time jobless claims fell more than forecast in the week ending July 5 to 346,000, a figure distorted by annual July shutdowns at auto plants....

Before the seasonal adjustment, claims rose 30,000 last week, Labor Department figures showed."

http://www.bloomberg.com/apps/news?pid=20601087&sid=av3OTgBo17is&refer=home

According to the actual numbers. Jobless claims rose 30,000. But after the government got through with the numbers. The headline became "Jobless Claims fall 58,000".

30,000 more people filed claims. And the government says, 58,000 less people filed claims. Why not just give the raw numbers and let people decide for themselves?

Is the government trying to create confusion among investors and stabilise the markets by increasing the randomness of investment decisions?

Answer 1

No doubt seasonal adjustment complicates the statistics, but it is very hard to argue that they are also distorted.  It is simply a fact that employment goes up and down in regular yearly patterns.  For example, retail businesses routinely hire temporary help during the Christmas holidays to assist with the rush.  These people are certain to get laid off in late December or early January.  Without a seasonal adjustment, the January numbers would show a big spike in unemployment, even though the jump is routine and expected.  What's the purpose of perpetually having to explain that an announced increase in unemployment is just seasonal when it's possible to accomplish the same thing by incorporating a seasonal adjustment in the report?

Answer 2

No, the government isn't doing this in any evil or conspiratorial sense.  And they *do* give us the raw numbers.

Employment has a strong seasonal pattern to it.  This is true both for jobs and for job-seekers (think of the seasonal impact of teaching jobs, resort jobs, and hordes of teenagers out of school, and many other things of that nature.)

So for many purposes, perhaps even most purposes, looking at seasonally-adjusted rates is better.   It smooths out *expected* and *known* increases and decreases due to the well-known time patterns.

But it is also true that for some OTHER purposes you may want to look at non-seasonally adjusted rates.  It's always good for the BLS and other agencies to give us the unadjusted figures at the same time as the adjusted ones.  The BLS *does* give us the unadjusted figures too.

Answer 3

If you don't like the adjusted numbers, you are free to ignore them.  The unadjusted numbers are available.

The question is, how do you interpret the raw numbers.  Normally, an increase in unemployment claims would indicate a problem in the economy.  But, in your own example, the auto industry lays off people in July regardless of economic conditions because this is when they shut down plants for maintenance and retooling.  Without seasonal adjustments, this would be a spike.  The other poster makes the same point about layoffs in January when retailers get rid of their Christmas help.

Answer 4

Actually, by using seasonally adjusted numbers, the claims and unemployment numbers are more easily compared across months.  For example, from November to December, many retailers hire folks for the Holiday season and then lay them off in January.  If you didn't adjust the numbers, you would get an exceptionally rosy picture of the economy which would be false in November, then a dismal picture of the economy in January, which would also be false.

In the case above, the shutdowns are a natural part of the auto industry, much like holiday hiring and not adjusting for them would skew the results.