Certain economic variables show a distinct pattern in their movement over a year. For instance, prices of vegetables will fall in winter season every year, sales will pick up in Christmas, diwali or any other festival season. These patterns, in general, can be decomposed into trend, cyclical and seasonal patterns. If any economic variable follows regular and predictable changes which persist every calendar year in a particular month or duration of months, then this type of pattern is known as seasonal pattern. The presence of such pattern is known as seasonality in time series. These are often short term, stable and predictable change that repeats over a one-year period.
The presence of seasonality in the economic variable can sometimes make it difficult to (a) identify the exact nature of the phenomenon represented by the sequence of observations, and (b) make any forecasts (i.e., predicting the future values of the economic variable). As such, it becomes imperative that the time series (i.e., the data of any economic variable over a certain period of years) is made free from any seasonality bias in the data. The process of removing this bias in the data is known as de-seasonalisation.
There are several econometric tools available to detect seasonality. Sometimes even a simple scatter plot can easily predict the existence of seasonal factors. On the other hand, ARIMA 12 (developed by census bureau of US), Census XI and Ratio to Moving Average are some of the methods by which seasonality factor can easily be removed from the series. In India, the Reserve Bank of India (RBI), releases every year in the month of September, seasonal factors for some of the important items, which are very helpful in de-seasonalising WPI and IIP series.