Virtually all headline news on the economy is based on seasonally adjusted statistics. So what does “seasonally adjusted” mean exactly?
The easiest way of explaining it is by thinking about Christmas. Retail sales are normally up in December from November – but to what extent is that due to Christmas being in December? Christmas is on the calendar at the same time every year, making it a perfect example of a normally occurring seasonal factor for retail sales.
Seasonal adjustment strips out normal seasonal fluctuations that otherwise affect data. It puts data on an equal footing so that data for any two months can be meaningfully compared to each other and to underlying economic fundaments like changes in interest rates or personal incomes.
Seasonally adjusted statistics result from raw data being fed into a complicated computer model. How the seasonal adjustment process works is the stuff of advanced studies in statistics (yawn inducing stuff to all but statisticians!) Suffice to say that the seasonal adjustment process takes into account not only Christmas, but also Easter (which can occur in either March or April), and trading days. The latter refers to the number of days in each month, and what day of the week any month starts on (which matters from the standpoint of how many business days there are, and whether any given business day normally results in more business than another business day). By stripping out those effects, all data are put on a level playing field.
CREA’s seasonally adjusted statistics on activity over MLS® Systems of Canadian real estate Boards and Associations are produced using the same model that Statistics Canada uses to seasonally adjust all of its data. CREA contracts with Statistics Canada to review the model annually, with the model tailored to each individual statistic for each Board and Association.
The effect of seasonal adjustment is that it reduces readings for months which normally get a seasonal lift, and boosts them for months which normally see a seasonal slump. Same goes for trading days: readings are reduced for months that have more trading days than others, and boosted for months with fewer trading days.
By looking at seasonally adjusted data, trends can be more easily identified compared to any attempt to do so using volatile raw data. Pretty important stuff if you want to know if things are getting, or are likely to get, better or worse!
