Trash seeps into the op-ed page of the WSJ in: “Seasonally Adjusted Jobs Numbers Offer Cold Comfort”:
The U.S. economy lost more than 2.7 million jobs between the middle of December and the middle of January, but the big news from the January jobs report was that the economy added 275,000 jobs during the same period.
And the reason:
Why the discrepancy? The Bureau of Labor Statistics touts “seasonally-adjusted” figures, which attempt to measure how recurring seasonal events affect employment. The raw figures are available to researchers, but the adjusted figures are the priority in public announcements.
Yet reporting a statistically adjusted figure as if it were original data is a mistake, and a significant distortion of reality that only adds to public distrust of the government and the media. People know that jobs were scarcer in January than in February, even if the government told them the opposite.
It´s “not as if it were the original data”. It´s a seasonally-adjusted data, and it´s not a “distortion of reality”. And why´s that? Just so that the ups & downs (noise) of the seasonal cycle don´t “cloud” the trend (signal). The chart shows a segment of the 1939 – 2015 period just to make the point.
Now, if the seasonal adjustment is being done properly, if you sum the raw (non seasonally-adjusted) data over the “seasons”, you should get the same result obtained if you sum-up the seasonally-adjusted data. And you do:
You should also not observe significant differences in the year on year growth rate of both the NSA and SA series. And that´s what you see:
Interestingly, however, there´s a predilection for presenting output growth data in “noisy” form, specifically as annualized rates of growth. As usual, the “noise” dampens the “signal”. The chart below shows the “imaginary” booms & busts in the annualized data.
“Imaginary” and distortive because it increases “anxiety” when all the while, for the last 5 years, the economy has grown at a monotonously low and stable 2.2% year on year!