A note on seasonal adjustments on Jobless Claims

[Zynx]

http://www.dol.gov/opa/media/press/eta/ui/current.htm

Last week, when seasonally adjusted jobless claims declined to 466,000 (initially reported), a lot of people here were skeptical because they said that actual claims increased. While that was true, we can see in the latest week why the non-seasonally adjusted claims are not used and why it is always better to use seasonally adjusted data. Non-seasonally adjusted jobless claims fell 78,000 from last week to this week. If you don't look at seasonally adjusted data, this sort of movement does not make sense. However, the seasonal adjustments create a much clearer picture with jobless claims, after revisions, falling from 462,000 to 457,000 this week.

In short, seasonal adjustments are not perfect, but they are not some government conspiracy either. Anyone who has had to deal with volatile time series data uses seasonal adjustments to obtain a clearer picture of what the actual trend is. Many seasonal patterns repeat from year to year and layoffs are one of those patterns. The DOL publishes the seasonal adjustment factors they use here: http://www.ows.doleta.gov/unemploy/claims.asp . Just look up the year you want to view, and you can see what the seasonal factor has been for any week over the history of the series. You can even see what they are going to use in the next few weeks.

Arguments against seasonal adjustments are almost always ill-informed and the arguments against seasonally adjusting jobless claims are no different.

[dmallind]

1. Damn you for bringing facts into a good ol' paranoid whine! NT

[Igel]

2. A parallel.

Climatologists don't look at absolute temperatures for much of their work because that makes for a very noisy signal. They look at abnormalities, deviation from the average for that day or time period. Of course it's colder in winter than in summer; the question is, is it colder or warmer in a given winter compared to the average winter. Of course Arctic ice declines in the summer--but is it the average, expected decline, greater, or less.

Then they compare departures from the norm.

Think of it as a transformation: The actual data is a mass of squiggles composed (via a good old composition function, if you will) of repeated, cyclical data and stuff that matters. So you yank out the repeated, cyclical bit. That just leaves the stuff that matters.

High school algebra, yes?

[Zynx]

3. Exactly. Thos who doubt seasonally adjusted economic data may as well

doubt the seasonally adjusted climate data. Hell, then they would be well on their way to being global warming deniers.