Exploring the safe haven properties of sukuk: a machine learning and K-cluster perspective

Purpose: This study aims to explore the safe-haven properties of sukuk and analyze the co-movement and interdependence between sukuk and conventional bond indices to provide insights into the potential role of diversification. Design/methodology/approach: The study uses the data set from 2012 to 2022, retrieved from the Eikon Reuter database. Different machine learning tools such as decision trees, random forests, gradient boosting and deep neural networks have been applied to capture the non-linear relationship and co-movement among the variables. Furthermore, K-clustering captures the hidden patterns and periods of high and low co-movements. Findings: The results state that the sukuk and conventional bond indices exhibit various degrees of co-movement influenced by regional and global market sentiments. The clustering analysis shows strong positive and negative correlations. The sukuk shows some instances of zero co-movement, but the results are inconsistent across all scenarios. Moreover, investors need to do their research first before investing in sukuk. Originality/value: This study uniquely applies K-clustering and advanced machine learning tools to understand the nonlinear relationship among variables better. In contrast, the previous studies mainly focused on linear relationships. It is critical to understand that financial variables tend to have nonlinear relationships, and these techniques best suit those needs.