Surprising Messages From a Mail Truck
March 10, 2015
Why We Can’t Avoid Bubbles
March 12, 2015
Last Friday we found out that the unemployment rate was 5.5 percent and 295,000 jobs were added to our labor force.
But they were not.
Actually the unemployment rate was 5.8 percent and we added 903,000 jobs. The difference is because of seasonal adjustment.
Where are we going? To how we can better grasp statistical headlines.
What is Seasonal Adjustment?
It is easiest to tell the story through department stores. During December 2014, there were 1.5 million department store jobs; in January, the number dropped to 1.36 million . But, was the economy contracting? No. It was all about holiday shopping. So, to avoid a misrepresentation of the state of the jobs market, government statisticians seasonally adjusted the numbers and wound up with almost identical totals. For December and January, department store jobs were close to 1.3428 million.
In addition to holiday employment pops, we also have the ups and downs of weather and summer lawns, school openings and school closings. As the BLS (Bureau of Labor Statistics) notes, by eliminating those types of blips, we can better see what really is happening to jobs and the economy.
The Debate
Not every one agrees that statistics should be seasonally adjusted.
In a recent WSJ column, a Harvard business professor suggested that the media convey both sets, the adjusted and non-adjusted numbers. Concerned that the economy depends less, for example, on malls because of online shopping, he says adjusted numbers that compensate for bad weather distort reality and policy decisions. For that reason, the unadjusted numbers should be more accessible and the adjusted formulas need to be up-dated.
Responding, a fivethirtyeight writer said that it is a mistake to focus on the weather. Rather it is recurring patterns that include the need for more tax preparers in March, landscapers during the summer, and fewer autoworkers during the new model prep shut down. Without numbers that recognize those patterns, the data would be misleading and obscure important trends. On the other hand, he concurs that the formulas could be tweaked.
Our Bottom Line: Do Statistics Lie?
When they are in spread sheets, employment reports and the GDP, numbers appear very believable. Perhaps we should be asking some questions about the jobs numbers in the headlines.
