Applying Approximate Entropy (ApEn) to Speculative Bubble in the Stock Market

Abstract

In contrast to the traditional duration dependence test, the article introduces an order statistic known as Approximate Entropy (ApEn) to investigate the presence of speculative bubbles for a cross-country sample. Using ApEn, the article examines four major crashes in the US, Japan, Hong Kong and India. In addition, the article also investigates the 1997 Asian crisis using weekly data for seven major Asian indices which include Hong Kong, Malaysia, Singapore, Korea, Taiwan, Indonesia and Japan. The results confirm that there are strong ‘tell-tale’ signs characterised by low ApEn level during many of these crash events. All the evidences using yearly as well as time series data (both discrete and rolling window analysis) point to a substantially lower level of ApEn during the crash. JEL Classification: G30, G01