Structure of Leading and Lagging Relationships between Early Warning Indicators' Detection of Financial Crises

This paper makes two proposals on the early warning indicators of financial crises. One of the proposals relates to the creation of early warning indicators. The current early warning indicators consist of economic indicators recorded at lengthy time intervals that are longer than monthly. For this reason, we cannot detect signs of uncertainty that lurk within the microscopic fluctuations such as stock prices. In this paper, we propose a new creation method of early warning indicators, based on the complex Hilbert principal component analysis (CHPCA) and the concept of renormalization in physics. Another proposal relates to the operation of early warning indicators. While there is research that claims that early warning indicators have stable leading/lagging structures, there are also studies that such structure does not exist because the type of financial crisis differs according to country and age. In this paper, we investigate the change of the leading/lagging structure of early warning indicators before and after the collapse of the Heisei bubble, chain collapse of financial institutions, and the economic shock caused by the collapse of Lehman Brothers using the CHPCA. In these financial crises, we clarify that there is no stable leading/lagging structure in the correlation between early warning indicators and discuss ways to reflect this result in monetary policy.