Volatility spillover effect between real estate investment trust and financial assets in South Africa

Abstract

This research examines the dynamics of volatility spillover e ects between the real estate investment trust (REITs) and nancial assets in South Africa. It examined the following nancial assets; bonds, stock, currency and commodities based on selected indices recorded using closing prices. Using the Vector autoregressive and Dynamic Conditional Correlation MGARCH (DCC-MGARCH) model on weekly data from May 2013 to November 2017, there is evidence of both volatility spillover e ects and correlation between assets. In appreciation of existing econometrics applications, a thorough examination of the methodologies available is done to understand how volatility is transmitted between markets and their interdependency from assets over stipulated periods. The dynamic conditional correlation model (DCC-GARCH) is used to examine correlation of variables over time. This research found that correlations are signi cant, and are both negative and positive dependant on the interaction to a particular asset. The assets are not independent of each other due to this their relation proves the existence of spillover e ects. Furthermore, this research sought to establish (based on literature) the degree of spillover e ects to each nancial asset. Results revealed there was a bidirectional volatility spillover between equity and bond markets. The presence of volatility spillover e ects con rms the e ciencies of cross market information transmission and integration of nancial markets of di erent economies. Similarly the e ect of bonds to stock is uni-directional and yield di erent results compared to the relation of stock to bonds in line with similar studies done on emerging economies including South Africa. Results were analysed to determine asset correlation and determine the degree of varying integration which is linked to investors risk premium expectations, cost of borrowing and diversi cation bene ts for investors. Certain class of assets stood out as they were negatively correlated to other assets. Currencies were positively correlated to commodities, and negatively correlated to stocks, real estate stocks and bonds. These results are considered to be good news for managers who are in the business of minimizing risk and examine the optimal portfolio allocations in line with their risk preferences. Based on the correlation, optimal portfolios can be composed using asset substitution that yields the greatest return and less risk. In this regard, negatively correlated securities could produce a diversi cation bene t which reduces risks. Furthermore, the decomposition of the variance of forecast errors over a period of 1-20 weeks as to whether an e ect of shock originates from REITS to other assets. Overall real estate accounts for 0.03 % to 53.43% of uctuations of the forecast error variance for other assets. Using the nancial 15 and industrial 25 as point of reference for stock index for the over examination of externalities; real estate a ects the returns of this index, accounting for 53.47% and while the Pula, palladium and British pound accounts for 3.52%,2.13% of the forecast error variances in for nancial index companies. While the nancials 15 index a ects the returns of other indices as well, contributing 29.68%, 11,60% for real estate and 1,74% for palladium. For Industrial 25, Financials 15 and Reits had the least e ects over time on its return with a decrease in uctuations at 4.08% and 0.56% respectively. Compared to the 1st week, during the 20th week the index returns are increasingly caused by shocks from other assets as its estimation of returns by itself decreases at 6.66%. Which means the returns of the index is not independent of other variables.