Understanding International Remittance Flows: A Comparative Analysis Across Distinct Corridors Using Hybrid Machine Learning Methodologies
Thesis title: | Understanding International Remittance Flows: A Comparative Analysis Across Distinct Corridors Using Hybrid Machine Learning Methodologies |
---|---|
Author: | Zhang, Jing |
Thesis type: | Diploma thesis |
Supervisor: | Musil, Petr |
Opponents: | Sixta, Jaroslav |
Thesis language: | English |
Abstract: | This study examines international remittance flows to/from ten diverse countries: Czech Republic, Philippines, Mexico, India, Ukraine, Vietnam, Poland, Slovakia, Turkey, and Bangladesh, during 2000-2024. A novel hybrid methodology, consisting of traditional economic analysis combined with machine learning techniques is applied, namely Temporal Pattern Recognition via Long Short-Term Memory (LSTM) networks and Determinant Identification through Extreme Gradient Boosting (XGBoost). Substantial determinant heterogeneity across countries at different levels of development is the major findings. While Real GDP is emerged to be the top determinant on a global level, the Czech Republic displays a unique profile, where the remittance-to-GDP ratio plays the top role, reflecting the unique transition profile of the Czech Republic among all countries in the world. It indicates that the Czech Republic has transitioned from being a remittance receiving country to a more bidirectional country since joining the European Union in 2004, with increasingly more determinant determinants. The temporal analysis reveals evolutionary patterns, i.e. determinants evolve in a systematic way with time, where economic factors become less important compared to institutional and integration factors. Moreover, the Czech Republic is more sensitive to economic shocks and exchange rate fluctuations compared to more mature sending-receiving countries for remittances. The study makes two principal contributions. Methodologically, it demonstrates that hybrid approaches, consisting of traditional econometric methods combined with machine learning, can be successfully applied to explain a complex phenomenon such as international remittance flows. Theoretically, it proposes a "Remittance Transition Model" to explain determinant evolutionary patterns of international remittance flows. Policy-wise, our findings imply that remittance policies should be designed according to the countries' development stages and transition positions, instead of a one-size-fits-all approach. |
Keywords: | Temporal Evolution; International Remittances; LSTM; Economic Integration; Macroeconomic Variables; Czech Republic; Socio-economic Factors; XGBoost |
Thesis title: | Understanding International Remittance Flows: A Comparative Analysis Across Distinct Corridors Using Hybrid Machine Learning Methodologies |
---|---|
Author: | Zhang, Jing |
Thesis type: | Diplomová práce |
Supervisor: | Musil, Petr |
Opponents: | Sixta, Jaroslav |
Thesis language: | English |
Abstract: | This study examines international remittance flows to/from ten diverse countries: Czech Republic, Philippines, Mexico, India, Ukraine, Vietnam, Poland, Slovakia, Turkey, and Bangladesh, during 2000-2024. A novel hybrid methodology, consisting of traditional economic analysis combined with machine learning techniques is applied, namely Temporal Pattern Recognition via Long Short-Term Memory (LSTM) networks and Determinant Identification through Extreme Gradient Boosting (XGBoost). Substantial determinant heterogeneity across countries at different levels of development is the major findings. While Real GDP is emerged to be the top determinant on a global level, the Czech Republic displays a unique profile, where the remittance-to-GDP ratio plays the top role, reflecting the unique transition profile of the Czech Republic among all countries in the world. It indicates that the Czech Republic has transitioned from being a remittance receiving country to a more bidirectional country since joining the European Union in 2004, with increasingly more determinant determinants. The temporal analysis reveals evolutionary patterns, i.e. determinants evolve in a systematic way with time, where economic factors become less important compared to institutional and integration factors. Moreover, the Czech Republic is more sensitive to economic shocks and exchange rate fluctuations compared to more mature sending-receiving countries for remittances. The study makes two principal contributions. Methodologically, it demonstrates that hybrid approaches, consisting of traditional econometric methods combined with machine learning, can be successfully applied to explain a complex phenomenon such as international remittance flows. Theoretically, it proposes a "Remittance Transition Model" to explain determinant evolutionary patterns of international remittance flows. Policy-wise, our findings imply that remittance policies should be designed according to the countries' development stages and transition positions, instead of a one-size-fits-all approach. |
Keywords: | International Remittances; Economic Integration; Macroeconomic Variables; Czech Republic; Socio-economic Factors; Temporal Evolution; LSTM; XGBoost |
Information about study
Study programme: | Economic Data Analysis/Official Statistics |
---|---|
Type of study programme: | Magisterský studijní program |
Assigned degree: | Ing. |
Institutions assigning academic degree: | Vysoká škola ekonomická v Praze |
Faculty: | Faculty of Informatics and Statistics |
Department: | Department of Economic Statistics |
Information on submission and defense
Date of assignment: | 19. 3. 2025 |
---|---|
Date of submission: | 24. 6. 2025 |
Date of defense: | 2025 |
Files for download
The files will be available after the defense of the thesis.