Seasonal adjustment

Seasonal adjustment is a statistical technique for the removal of seasonal variation from a time series, usually of quarterly or monthly data.

The conceptual idea is to represent the series as a combination of three components: Trend (C), Seasonality (S) and Irregular (I). Trend here means a smooth series, regarded as the underlying level of the series, not a change in level. The letter C is used because the trend may include the economic cycle. The seasonal component is removed, leaving the trend and irregular. Note that the seasonally adjusted series may still be fairly rough if there is much irregularity in the original series.

Any series may be reduced to stability by removing a very smooth (often monotonic) trend and a stable or very slowly changing seasonality. The variability in the residual stable series will include short-term trend movements, rapid changes in seasonality and the irregular; the art of good seasonal adjustment is to allocate this variability correctly.

The commonest technique for seasonal adjustment is X-11 and its successors.

Copied from Wikinfo under GFDL.