Fuzzy gravity approach for determinants of exports

Purpose: The present study proposes the fuzzy gravity model approach to examine the main factors that affect the export volume of Turkey and whether the European Union Customs Union (EUCU) membership affected the export of it. Design/Methodology/Approach: A fuzzy approach was developed for the gravity model by using these variables. The dependent variable was the export volume of Turkey to 204 countries throughout the whole world, and the explanatory variables were the gross domestic product (GDP) of the countries that are subject to export and their distances to Turkey (dij), populations, whether the seaway transport was possible, and the membership to the EUCU. The last two items were added as dummy variables in the model. Findings: The results showed that gross domestic product (GDP) of the country which Turkey exports to, its being a member of the EUCU, and its population affected the export volume of Turkey positively. However, the distance of the target country to Turkey had a negative effect. The coefficient of distance variable was found to be negative fuzzy and ones of the seaway was fully fuzzy. However, when seaway variable was taken together with distance, it was positive fuzzy, which showed the positive effect of the opportunity of transportation by sea in long distances on export volume. Practical Implications: It is thought that the policies to be established based on these findings will be beneficial in improving the country's exports. Other available fuzzy regression approaches may be tried in future studies to obtain a fuzzy model. In addition, the import or foreign trade structure of Turkey may be dealt with the same approach. Originality/Value: In this study the fuzzy gravity model approach was used as a novelty, which was different from the methods in the literature, because the gravity model is criticized sometimes that its theoretical basis is weak and the relations between dependent variables and explanatory variables are not adequately clear.peer-reviewe