Why is housing data reported differently in terms of `original’, `trend estimate’ or `seasonally adjusted’ figures?
Which should we pay attention to?
According to definitions offered by the Australian Bureau of Statistics, that probably just depends on what you want to see.
`Original’ means that very little has been done to the data. It is raw data, straight from the survey and shows all the ups and downs of the data being measured. For instance, there will invariably be a spike in retail sales in December each year, due to increased spending for Christmas.
`Seasonal adjustment’ allows for and removes the regular, recurring influences that could distort the short-term view of what is happening. That means the December spike in retail spending is smoothed out.
A `trend estimate’ has not only had its seasonal factors allowed for, but it has also had effects from once-off or irregular events removed, so that the data is not thrown off by what are essentially random events. For example, the Sydney Olympics distorted the ABS statistics on non-residential building approvals for that year.
Trends are usually referred to in terms of direction, for example whether the long-term pattern of behaviour is showing an increase or a decrease.