The ABS explains “trend estimates”
I've argued several times that the broadcast media has misrepresented quarterly unemployment by focusing on (the scarier of either) the raw or seasonally-adjusted data. Although there's a great deal to be said for just eyballing a trend when you have a long time-series that accurately samples an 'atomic' phenomenon, trends in a series with statistically simple ('normally distributed') variability, like unemployment, can reasonably be represented by a smoothed trend estimate.
"Trend estimates are obtained by smoothing the seasonally adjusted estimates, with an assumption that the irregular component is random and normally distributed. Distortion of the trend estimate will occur in the presence of an unusual event, and if no correction for the impact is introduced, then the trend estimate can be misleading." Extract from 1350.0 - Australian Economic Indicators, Aug 2009
As the ABS release argues, this procedure goes wrong when it's no longer 'business as usual.' But the assumption that the world today is just like the world yesterday is both prudent and pretty successful, especially at the leading edge of a trend with established variation. Correlation of smoothed data series is a different matter: it may be a species of fraud.
Posted on 07/31 at 04:00 PM.
Your comments