Forecasting Foreign Direct Investment by Using Bass Diffusion Model Integrated with Genetic Algorithms

Abstract

The evolution of industrial clusters is a critical factor in the strategic development of locations for high-tech industries. Most previous studies have seldom quantified the location selections of Taiwanese IC design firms engaging in foreign direct investments in China because their access to crucial data may have been limited. This work developed a novel diffusion model to illustrate the extension of IC design clusters from Taiwan to China and chose foreign direct investment (FDI) as a quantitative indicator to measure the size of Taiwanese IC design industrial clusters in different Chinese regions. This study aims to understand what the distinctions in the process of FDI implementation are among the Taiwanese IC design companies which choose different Chinese regions to engage in FDIs. We also modified the conventional Bass model by optimizing parameters using genetic algorithm (GA) in conjunction with nonlinear least square method. The simulation is iterated 3,000 times. Finally, t-statistics were used to compare clustering features between the eastern and southern China areas. Simulation results demonstrate negligible standard deviation in the optimized parameters, which confirms the reliability of our findings. Furthermore, comparison results demonstrate that the Bass model optimized by GA integrated with NLS approach is more stable and accurate than the Bass model optimized by GA. The proposed approach is applicable to other high-tech industries and other locations around the world.