Backward recalculation of seasonal series affected by economic crisis: a Model-Based-Link method for the case of Turkish GDP

When attempting to deal with the recalculation process, it is hard to answer the question “Does the recalculated series include economic events and seasonal behaviours in the past?”. This paper discusses some alternative backward recalculation methods and presents the applications and their results relative to the Turkish Gross Domestic Product (GDP) series. Using comparative analysis, it is shown that ordinary ARIMA forecasts and signal extraction methods are not successful at taking into account past events in the backward recalculated series. A new innovative method, named Modelbased-link, is then proposed and suggested by the authors in order to be able to take past economic events and seasonal patterns into account when the series is to be backward recalculated. A first application of this new method is run on the quarterly series of the Turkish GDP. In addition, it is shown that the Model-based-link method can be extended to data sets of different frequencies (i.e. annual data). Consequently, it can be claimed that a comparable recalculated quarterly and annual Turkish GDP series for forthcoming data is obtained. The paper is structured as following: section 1 introduces the reader to the state of the art in the current literature; section 2 defines the information set to be backward recalculated and presents some statistics on the data while section 3 presents the main methodological statistical aspects of classical methods compared to the methodological scheme of the Model-based-link that can be used for the recalculation process. Section 4 presents results of the methods mentioned in the previous section and section 5 discusses the extension of the Model-based-link method to monthly data and includes an application for annual data; section 6 concludes. Finally, section 7 presents topics for discussion and challenges for continuation of the analysis.