B1662
Title: Modelling seasonality with long seasonal period
Authors: Jiri Prochazka - University of Economics, Prague (Czech Republic) [presenting]
Samuel Flimmel - University of Economics in Prague (Czech Republic)
Abstract: The main aim is the discussion of different methods used for modeling seasonality with long seasonal periods. A special focus is given on procedures based on the representation of the seasonal pattern using basis functions. Advantages and disadvantages of the various methods are discussed. Specifically, the most commonly used methods for modeling seasonality with short seasonal periods are shown not to be appropriate in the case of long seasonal periods since they can suffer from overparametrization if a sufficient number of observations is not available for the model estimation. On the other hand, the decomposition of the seasonal pattern using different types of basis functions is argued to provide a potential remedy for modeling seasonality with long seasonal periods in the case of a small number of observations. The study is supported with illustrations that make use of simulated as well as real-life data sets.
Journal