South African Sector Return Correlations: using DCC and ADCC Multivariate GARCH techniques to uncover the underlying dynamics

This paper explores the dynamics of return co-movements between the largest economic sectors in South Africa, specifically with a view to shed light on the inter-sector diversification potential of domestic investors over time. It has been widely documented that investors have a home-bias when it comes to investing, and as such may be exposed to periods of increased co-movement between assets held locally across different sectors in their portfolios. Such periods of increased homogeneity in the movement of asset prices negate the benefits from diversification within the domestic financial market. The paper utilizes Dynamic Conditional Correlation (DCC) and Asymmetric-DCC Multivariate Generalized Autoregressive Conditional Heteroskedasticity (MV-GARCH) techniques to isolate the time-varying conditional correlations from the conditional variance component. These series are then used to study whether changes in market conditions and overall sentiment influence the dynamics and aggregate level of co-movement between sectors. The results firstly suggest that using static measures of historic co-movement between asset returns across sectors in order to evaluate a portfolio’s diversification potential are inaccurate. Significant leverage effects are also found in the dynamics of co-movement between the sector pairs, with negative shocks being followed in all cases by higher aggregate levels of co-movement. The results also suggest that periods of heightened global- and domestic market uncertainty magnifies the co-movements between sectors and in so doing undermines the ability of investors to diversify across local sectors.