Abstract
In light of the recent global financial crisis, the questions about financial bubbles and its detection has come back to the stage, attracting concerns from academics and policy makers. Evans(1991) created the famous periodically collapsing bubble model, which has been widely used as data generating process in previous bubble detection research. Since Diba and Grossman (1988) started the Bhargava Test for bubbles detection, the academic study of bubbles has gradually expanded to find more accurate and general method for detecting bubbles. Homm and Breitung (2012) put forward several modified statistics for bubble detection and compared their performance. Phillips, Yu et al.(2014) proposed the famous PSY method which has excellent performance in real-time bubble monitoring. During this research, we developed a new bubble model and two data generating processes which contain several meaningful properties and overcome the improper assumption of previous data generating mechanism. Then, our paper focus on comparing the performance of different bubble detection methods under three bubble data generating processes in a comprehensive view.