Seasonally Adjusted Data
Dr. Electroid builds a new weather control device. As part of his evil plan for world domination, he dumps 50 inches of snow on Phoenix. We’re not sure how world domination follows from a big snowstorm in Phoenix, but uh...we’re sure Dr. Electroid knows what he’s doing. Meanwhile, Captain Condor flies from his secret superhero lair high up in the Sierra Nevada mountains. He captures Dr. Electroid, destroys the weather device, and saves the day.
Cut to...you. You're a researcher for the national weather service.You're locked in a basement of the Washington DC headquarters, with no windows and no other people, surrounded only by the blinking lights of weather measurement equipment. You don't have any cell phone reception or Internet access. All you know is that your equipment shows 50 inches of snow in Phoenix. You call your boss: “Phoenix is the snowiest place in the nation. Call the press! Get started on your game-changing academic paper on climate change!”
Uh, well...wait a minute. The 50 inches of snow doesn't indicate a real change in Phoenix's long-term snowfall. It's just a one-time event (in this case, caused by a supervillain). The lesson: you can't always trust raw data. Numbers need context. The need for context leads us to seasonally adjusted numbers. That is, the numbers get massaged to take into account the natural impact of the time of year.
For instance, the unemployment rate. This stat gets reported on a seasonally adjusted basis. Every year, going into Christmas, companies hire additional seasonal workers. Package delivery people need extra drivers. Amazon and other retailers need extra people to process orders. Malls need elves to put crying kids onto the laps of Santas. Suicide hotlines need extra people to take distressed calls from family dinners.
On an unadjusted basis, the unemployment rate would always decline headed into the holiday season. All these extra workers would lead to lower unemployment. Every year. But eventually, Christmas comes and Christmas goes. Everyone opens their presents and stuffs everything into closets for potential regifting next year. All the extra holiday workers are back on unemployment...at least until spring, when they can start lining up for summer gigs. Maybe jobs as lifeguards and stand-up comics on cruise ships.
So going into the holidays, every year would see a drop in unemployment with unadjusted data. And then, after the holidays, there would always be a rise in the unemployment rate. If the figure was reported this way, it would be hard to read. Much of the movement during this time of year would be a result of seasonal factors. The unadjusted statistics wouldn’t mean much as an economic indicator. It would be hard to see what's really going on in the actual economy, aside from the usual ticks up and down. Is this year better than last year? How does this year stack up historically? How's the labor market really doing? It would be hard to tell. Much of the data would represent seasonal noise.
So economists strip out the seasonal movements. They make guesses based on what’s happened before about what the data should do as a result of these seasonal factors. They then take these seasonal moves out, and only report the changes that don’t have to do with seasonal workers.
Theoretically, this seasonally adjusted data doesn't include the moves in unemployment that come from the holiday staffing. Instead, the numbers track the underlying health of the labor market…like they’re supposed to do. Of course, if you spend your winters as a mall Santa, your springs dressed like the Easter Bunny, your summers as a snow cone salesperson, and your autumns as a jack-o-lantern carver, you might never get counted in the seasonally adjusted numbers.