Causality and Correlations between BSE and NYSE indexes: A Janus Faced Relationship

TL;DR

This study analyzes the multi-scale correlations and causality between NYSE and BSE indexes over 300 months, revealing complex phase-locking, non-stationary modulations, and a bidirectional causality that highlights the influence of smaller markets on larger ones.

Contribution

It provides a detailed multi-scale analysis of NYSE and BSE index relationships, uncovering phase-locking during turbulent times and reverse causality at high frequencies, which is a novel insight.

Findings

NYSE Granger-causes BSE with a 9-month lag

BSE influences NYSE at high frequency with a shorter lag

In-phase relationships are prominent during turbulent periods

Abstract

We study the multi-scale temporal correlations and causality connections between the New York Stock Exchange (NYSE) and Bombay Stock Exchange (BSE) monthly average closing price indexes for a period of 300 months, encompassing the time period of the liberalisation of the Indian economy and its gradual global exposure. In multi-scale analysis; clearly identifiable 1, 2 and 3 year non-stationary periodic modulations in NYSE and BSE have been observed, with NYSE commensurating changes in BSE at 3 years scale. Interestingly, at one year time scale, the two exchanges are phase locked only during the turbulent times, while at the scale of three year, in-phase nature is observed for a much longer time frame. The two year time period, having characteristics of both one and three year variations, acts as the transition regime. The normalised NYSE's stock value is found to Granger cause those of BSE, with a time lag of 9 months. Surprisingly, observed Granger causality of high frequency variations reveals BSE behaviour getting reflected in the NYSE index fluctuations, after a smaller time lag. This Janus faced relationship, shows that smaller stock exchanges may provide a natural setting for simulating market fluctuations of much bigger exchanges. This possibly arises due to the fact that high frequency fluctuations form an universal part of the financial time series, and are expected to exhibit similar characteristics in open market economies.