Tuesday, June 19, 2012

Skewed Seasonal Adjustment?

Today we have a bit more detail on our recent conversations about employment data.

Did the Great Recession throw a wrench into the seasonal adjustment process? Some U.S. economic observers think the headline payroll employment numbers were overstated for January through March due to a seasonal adjustment quirk. The Bureau of Labor Statistics updates the factors used for seasonal adjustment each year (see previous post with details on seasonal adjustment and why we use it). On June 6th, in The Wall Street Journal’s Fed watch article, Jon Hilsenrath states, “fed officials also worry that data might be skewed by seasonal adjustments that statisticians use to smooth normal monthly swings in the data.” 

To test this theory with Oregon data, our resident Current Employment Statistics guru David Cooke plugged in the seasonal adjustment factors we used back in the latest “normal” year to see what the difference was compared with our current factors. I used the old factors that were derived from data ending in December 2007.

This test for Oregon data didn’t support the skewed theory mentioned above. Rather, the 2007 seasonal adjustment factors, when applied to the 2012 employment estimates, produced the opposite effect. The older seasonal factors resulted in higher employment estimates for January and February, by about 1,700 jobs or 0.1 percent. The subsequent two months came in lower by an average of 900.

Other findings of this little test: both in manufacturing and in construction, the old and the new seasonal adjustment factors produced about the same results. On the other hand, retail trade was noteworthy in that it has experienced diminished seasonality compared with four years ago.

Testing the theory of skewed seasonal factors for the case of Oregon doesn’t provide irrefutable proof. Rather, this test shows that things didn’t work out as proposed for this one set of circumstances. But for Oregon, at least, it appears that relatively weak employment reports this spring can’t be blamed on a quirk in the seasonal adjustment process.

Want more information on seasonal adjustment, the factors used to make adjustments, or Oregon's monthly payroll employment numbers? Contact David Cooke!

No comments: