Why We Use “Seasonally Adjusted” Data

This graph from Calculated Risk is a great example of why we use “seasonally adjusted data”:

Retail Employment, Jan 98-Jan 11, Seasonal and Non-Seasonal Adjusted

The red line is the raw, unadjusted monthly data.  It’s noisy. It goes up and down dramatically.  But it has a pattern to the rises.  Every October, November, and December it rises dramatically and then falls in January.  That’s the holiday rise in retail employment – all those clerks and salespeople pushing us to buy, buy, buy for Christmas.  But since we know that hiring always rises in the the last 3 months, it’s not really telling us anything about the longer term trend.  So, through the magic of seasonal adjustment statistical methods we can take out the predictable seasonality.  What’s left is the blue line:  the underlying trend.  That’s more informative if you’re wondering “What’s the economy doing now?”