Modelling delayed correlation between interest rates and equity market returns

Abstract

This study models the interaction between interest rates and equity markets using wavelet analysis. This approach facilitates assessment of the lead-lag relationships in an intuitive way considering variation across frequencies and over time. Analysis is done progressively on varying scales where the lower scales encompasses high frequency components of the data over shorter time scales whereas higher scales encompass low frequency components over longer time scales. The study uses daily data obtained from Kenya for the period October 2003 to October 2019. The Nairobi Securities Exchange 20 share index returns are used as a proxy for equity returns whereas the interbank rates represent interest rates. Three key findings emerge: (1) There is at least two months delay in the correlation between interest rates and equity market returns, (2) The correlation is lower (correlation coefficients of 0.3 and below, including negative correlation coefficients) in the lower time scales of 4 to 8 days and higher (correlation coefficient of 0.3 and above) in higher time scales of 512 to 1,024 days, and (3) Equity returns lead interest rates in Kenya during the period of study. Unlike common practice of assuming joint stationarity of variables, the findings reiterate the need for modelling dynamic relationships considering delays, time variation and scaling over time horizons.Investors, portfolio managers and policy makers alike should therefore be cognisant of the dynamic nature of the relationship between interest rates and asset markets bearing in mind that changes in one variable may not have immediate impact on the other.