Risk parity as an asset allocation technique: evidence from South African capital markets

Abstract

This paper investigates the asset allocation technique known as Risk Parity, whereby assets are allocated such that they contribute equal amounts of risk to the overall risk of the portfolio. It is a relatively new technique and one which has grown in popularity and stature amongst Hedge Fund and asset managers alike. Academics have recently also come to fore, documenting the flaws with the mean-variance framework and have begun looking towards portfolio construction techniques based solely on predicted risk, not expected return. The prior literature on the topic is exclusively done abroad and finds that an unlevered Risk Parity portfolio, despite being inferior to other portfolios from a return perspective, is superior in terms of its risk-adjusted return, or Sharpe Ratio. Many academics propose the idea of levering up the Risk Parity portfolio so that its standard deviation matches that of another, riskier portfolio. This paper analyses five portfolio strategies, namely an unlevered and levered Risk Parity, a traditional 60/40, a minimum variance and a maximum Sharpe Ratio (tangency) portfolio. The first part of the paper will categorise the equity asset class into the FINDI and RESI indices, as well as using the ALBI and South African Property Index (PROP). The second part of this paper, Test 2, will subcategorize the equity portion of the strategies into Value and Momentum, using style indices, thus testing for evidence of style anomalies on the JSE. It will still use the ALBI and PROP as the other asset classes. The key findings are that for both tests, the unlevered and levered Risk Parity strategies underperform the FTSE/ALSI benchmark over the sample period, concluding that at the given level of risk free rates, Risk Parity as an asset allocation technique is not superior. Furthermore, Test 2 provides superior annualized returns for all but one of the strategies, indicating the possibility of style anomalies on the JSE.