<abstract xmlns="http://www.w3.org/1999/xhtml"><p>In the most developed countries the first estimations of Gross Domestic Product (GDP) are available 30 days after the end of the reference quarter. In this paper, possibilities of creating an econometric model for making short-term forecasts of GDP in B&amp;H have been explored. The database consists of more than 100 daily, monthly and quarterly time series for the period 2006q1-2016q4. The aim of this study was to estimate and validate different factor models. Due to the length limit of the series, the factor analysis included 12 time series which had a correlation coefficient with a quarterly GDP at the absolute value greater than 0.8. The principal component analysis (PCA) and the orthogonal varimax rotation of the initial solution were applied. Three principal components are extracted from the set of the series, thus together accounting for 73.34% of the total variability of the given set of series. The final choice of the model for forecasting quarterly B&amp;H GDP was selected based on a comparative analysis of the predictive efficiency of the analysed models for the in-sample period and for the out-of-sample period. The unbiasedness and efficiency of individual forecasts were tested using the Mincer-Zarnowitz regression, while a comparison of the accuracy of forecast of two models was tested by the Diebold-Mariano test. We have examined the justification of a combination of two forecasts using the Granger-Ramanathan regression. A factor model involving three factors has shown to be the most efficient factor model for forecasting quarterly B&amp;H GDP.</p></abstract>

In the most developed countries the first estimations of Gross Domestic Product (GDP) are available 30 days after the end of the reference quarter. In this paper, possibilities of creating an econometric model for making short-term forecasts of GDP in B&H have been explored. The database consists of more than 100 daily, monthly and quarterly time series for the period 2006q1-2016q4. The aim of this study was to estimate and validate different factor models. Due to the length limit of the series, the factor analysis included 12 time series which had a correlation coefficient with a quarterly GDP at the absolute value greater than 0.8. The principal component analysis (PCA) and the orthogonal varimax rotation of the initial solution were applied. Three principal components are extracted from the set of the series, thus together accounting for 73.34% of the total variability of the given set of series. The final choice of the model for forecasting quarterly B&H GDP was selected based on a comparative analysis of the predictive efficiency of the analysed models for the in-sample period and for the out-of-sample period. The unbiasedness and efficiency of individual forecasts were tested using the Mincer-Zarnowitz regression, while a comparison of the accuracy of forecast of two models was tested by the Diebold-Mariano test. We have examined the justification of a combination of two forecasts using the Granger-Ramanathan regression. A factor model involving three factors has shown to be the most efficient factor model for forecasting quarterly B&H GDP.

Modelling and forecasting GDP using factor model: An empirical study from Bosnia and Herzegovina

Croatian Review of Economic, Business and Social Statistics

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

In the most developed countries the first estimations of Gross Domestic Product (GDP) are available 30 days after the end of the reference quarter. In this...

Modelling and forecasting GDP using factor model: An empirical study from Bosnia and Herzegovina

Published Online: Jun 08, 2020

Page range: 10 - 26

Received: Dec 23, 2019

Accepted: May 12, 2020

DOI: https://doi.org/10.2478/crebss-2020-0002

Keywords
efficiency, factor model, Gross Domestic Product of Bosnia and Hercegovina, unbiasedness

© 2020 Ademir Abdić et al., published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Modelling and forecasting GDP using factor model: An empirical study from Bosnia and Herzegovina

Published Online: Jun 08, 2020

Page range: 10 - 26

Received: Dec 23, 2019

Accepted: May 12, 2020

DOI: https://doi.org/10.2478/crebss-2020-0002

Keywordsefficiency, factor model, Gross Domestic Product of Bosnia and Hercegovina, unbiasedness

© 2020 Ademir Abdić et al., published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Keywords
efficiency, factor model, Gross Domestic Product of Bosnia and Hercegovina, unbiasedness