Understand Data Basics

Components of Seasonal Adjustment

Seasonal adjustment is a process of using analytical techniques to estimate and remove seasonal and calendar effects, which may conceal and distort the true underlying movement of an economic time series. 

The seasonally adjusted data series facilitate a better assessment of their recent movements, including the timelier identification of turning points. According to the decomposition theory, every time series comprises four components:

• Trend – growth or decline observed over an extended period of time.
• Cycle – sinusoidal fluctuation around the trend, influenced by economic expansions and contractions.
• Seasonal – intra-year periodic variation that repeats itself every year.
• Irregular – short-term erratic random fluctuations, caused by unanticipated events.

Currently, DOS uses the X-12 ARIMA procedure developed by the US Census Bureau, in carrying out seasonal adjustments. This procedure is widely used among national statistical offices around the world. 

Seasonally Adjusted (SA) Data vs Non-seasonally Adjusted (NSA) Data

Non-seasonally Adjusted (NSA) data reflects the actual economic events that have occurred, while Seasonally Adjusted (SA) data represents an analytical elaboration of the data designed to show the underlying movements that may be hidden by the seasonal variation. 

SA data is particularly useful during instances where NSA data contains strong seasonal patterns and hinders detailed in-depth data analysis. In this case, NSA data period-on-period (e.g., month-on-month or quarter-on-quarter) growths may be masked by seasonal fluctuations and only year-on-year growths, which are not as sensitive in detecting short-term changes in growth momentum, may be quoted for reporting purposes.

It is possible that the non-seasonally adjusted (NSA) data shows an increase over the previous month but the seasonally adjusted (SA) data shows a decrease, when the increase is less than the usual seasonal increase. To illustrate, consider the retail sales index (RSI) which tends to increase sharply in December during the festive season. Suppose that an economic downturn results in a weak performance of the retail sector, the unadjusted RSI for December might still show a moderate increase over November. But because this increase is lower than that for a typical December, the seasonally adjusted RSI for December would be lower than the corresponding seasonally adjusted RSI for November.

For more information on seasonal adjustments of time series, you may refer to the Information Paper on Seasonal Adjustment of Economic Time Series. [PDF, 239.6 KB] You could also refer to the article on Seasonal Adjustments of Time Series [PDF, 32.4 KB].